Barry Kauler

Woof uploaded, February 8, 2010

There have been three commits since the previous upload of Woof:

Quirky: experiment use mostly busybox 1.16.0 applets not good, mostly reverted
dir2pet, petspec scripts improved
Puppy 4.3.2-experiment1 built with this release of Woof

Grab this one if you are interested in building your own Puppy in the 4.x series.

Woof home page:
http://bkhome.org/woof/

Experimental unofficial Puppy 4.3.2

Here it is again, with the fix for ROX-Filer:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/test/puppy-4.3.2-experimental1/

The ISO has all the modem drivers in it, which is why it is a bit big (112MB). SCSI drivers are missing.

There is a forum thread for feedback:

http://murga-linux.com/puppy/viewtopic.php?t=52280

I'm only really interested in bug reports of a certain type: if something worked in 4.3.1 but not in this 4.3.2 build. It would be nice to know about the reverse also: something that did not work in 4.3.1 and does in 4.3.2.

I must reiterate that I did not create this with the intention that it will be the next official 4.3.2 release. I'm mostly just wanting to know that Woof builds a sane Puppy 4.x.

Improved modem scripts

Rerwin (Richard) has posted packages for improved modem detection and management:

http://www.murga-linux.com/puppy/viewtopic.php?p=388235#388235

He has commented:

Barry and technosaurus might wait a while before integrating them, to allow users to "shake it out"

So, everyone running Puppy 4.3.1 or later variant, is invited to test these packages!

ROX-Filer: fix for focus bug

I described this problem here:

http://bkhome.org/blog/?viewDetailed=01387

I would like to thank 'perthie' who responded, informing me of a fix posted to the Dpup forum thread.

I found it:

http://www.murga-linux.com/puppy/viewtopic.php?t=50728&start=15

...thanks to 'rcrsn51' for the fix.

I have tested in pup 432 and the fix works. I have implemented this fix in Woof, adding this code to /etc/profile:

#100208 rox-filer needs this for GTK >= 2.18, fixes focus problem...

if [ `grep '^gtk+-2' /root/.packages/woof-installed-packages | head -n 1 | cut -f 2 -d '.'` -ge 18 ];then

 export GDK_NATIVE_WINDOWS=true

fi

Gecko: Adobe Flash player crashes

With my "432" pup, which has Adobe Flashplayer 10.x, it crashes when I try to play a video at youtube.com. In /tmp/xerr.log I see "illegal instruction".

So I tried Flashplayer9, same thing.

So I tried pskin's 431 pup, same thing.

Then I tried WattOS, and that works. WattOS has the Adobe player.

Just on the off-chance, I got the player out of WattOS and tried it in pup 431, still crashes.

So, it's something else. I suspect the MTRR registers. The kernel used in WattOS has MTRR disabled. I suppose that I could try that, recompile the kernel for Puppy with MTRR disabled.

Experimental Puppy 4.3.2

I have built this using Woof, so it has the latest infrastructure, that is, all the underlying scripts. I also updated or added these packages:

915resolution, bones, cpu-scaling-ondemand, flashplayer10, grub4dos (instead of grub), lamesmbxplorer (instead of pnethood), ntfs-backup, pburn, pcd, pfind, pmusic (instead of aqualung), precord, pupradio, seamonkey_addone_flashblock, seamonkey_addon_user_agent_switcher, sfs-converter, util-macros, woo-ff, floppy-format

I am using it right now, and it is working fine. There is just one thing, a bug with ROX-Filer that plagued me awhile back with Quirky. When I built Quirky with GTK 2.18.3, ROX-Filer exhibited a problem in which file icons would stay shadowed/selected. For example, if I right-click on a file then choose to open it in the text editor, the file icon remains shadowed, even after the text editor is closed. It is effectively still "selected" and this has an adverse effect if you don't click in the window to de-select it, as if you for example click on some other file, the previous one is actually chosen -- awful!

In Quirky, I rolled back to GTK 2.16.6 and that fixed it. However, to my great surprise, I now have the problem in 4.3.2. I was thinking "oh no!!!!", but then I discovered that Woof built Puppy 432 with my latest GTK 2.18.3 packages. So, bear in mind that the ISO I have uploaded has the "wrong" version of GTK, which may also adversely affect some applications.

I'll do another build tomorrow.

Am I planning to release an official Puppy 4.3.2? No, I built this because the 'rdc' Xorg driver does not work in Xorg 7.5, which is what Quirky uses. I have installed 432 into the Gecko Edubook, but it turns out that the 'rdc' driver is too buggy so I'm only using the 'vesa' driver -- so I might as well have stayed with Quirky.

...so that's how 432 came into being. As I have come this far, I'll fix the ROX-Filer problem, and might even recompile the kernel to add the elantech touchpad support that Jemimah wanted.

Anyway, if you feel like you want to play with it, get Puppy "unofficial" 4.3.2 with "wrong" GTK version here:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/test/experimental-432/

BUT, I must emphasise that some apps may be adversely affected by this later GTK. So please do not report that Abiword worked in 431 but has become strange in 432. If in doubt, wait and I'll upload 432 with the correct GTK in a day or two.

Note, after I have built 432 with correct version of GTK, I'll upload Woof that has been used to build this pup.

2.6.30.5 kernel recompiled

Forum member pskin has already done this, but I have also been through the exercise as I wanted to make some other minor configuration changes.

I patched the kernel source to support audio on the Gecko Edubook, and made these configuration changes (relative to the kernel used in 4.3.1):

Kernel compression mode (gzip) -- changed from lzmo
(15) Kernel log buffer size -- up from 14
apm bios support: [*] ignore user suspend

...after compiling the kernel, I remembered that Jemimah wanted support for the Elantech touchpad -- doh! -- well, maybe I'll compile it again soon.

The changes are very small, and all existing modules created for Puppy 4.3.1 should still work.

I have uploaded the patched source and the .config file:

http://bkhome.org/sources/kernel-2.6.30.5-pup432/

'tr' mystery

One of the scripts in Woof has suddenly gone crazy. I found the cause: the 'tr' utility has suddenly developed an abnormality. The script uses 'tr' to convert all lower-case letters to upper, like this:

# echo 'abc' | tr [a-z] [A-Z]
ABC

...well, the output used to be 'ABC', but now it has become 'abc'. In other words, tr is just passing the characters through unchanged:

# echo 'abc' | tr [a-z] [A-Z]
abc

I tried with the latest 1.16.0 'busybox' executable, same problem.

I rebooted with "pfix=ram" and 'tr' works. I ran my Woof script and 'tr' becomes broken. I have used this '2createpackages' script many times without this problem. I am so puzzled.

Ah, if I open a new terminal, 'tr' works again. More specifically, if I open another terminal in the same Woof directory as my script, 'tr' is broken. Open a terminal anywhere else, 'tr' works.

However, if I do it this way, tr comes good again:

# echo 'abc' | tr '[a-z]' '[A-Z]'
ABC

I put the single-quotes into my script, and it is now working. Yeah, but why?????

Gecko: DMP X-Linux

I have installed X-Linux, which is provided by DMP, the manufacturers of the Xcore86 CPU and the Gecko Edubook.

It's a simple console-only distribution (no X).

I installed it into sda2 of the SD card and made an entry in menu.lst. It booted okay.

I especially wanted to know if it would power-off the Gecko. Well, guess what, it doesn't!!!!! I typed "poweroff" at the prompt, it did the normal shutting down things, got down to "System halted" and that was it, no power-off.

But, I don't know how the software request to power-off actually works. Perhaps the fault is not on the actual circuit board but in the design of the power supply -- just guessing.

If DMP themselves can't do it, what can I say? I'm not going to rush into writing what I'm thinking. I'll keep investigating, and will post my conclusions on the Edubook in a couple of days.

Gecko: success booting FreeDOS

Yay, I finally got there!

Thanks for all the feedback guys.

What I ended up doing was using GRUB4DOS. I intended to put in either GRUB or GRUB4DOS anyway, to offer dual booting between the DOS shell and Puppy/Quirky.

Using GRUB4DOS, it was so easy. I booted Quirky 006 and used shinobar's package (which is already in Quirky) to install to the SD card, sda1, which is the little 256MB fat32 partition.

I then edited 'menu.lst':

title Boot FreeDOS 1.0 in sda1
root (hd0,0)
chainloader /kernel.sys

I then placed 'kernel.sys' and 'command.com' from the FreeDOS live-CD into sda1.

Rebooted, it works!

I then placed the 'spiflash.exe' utility and the 'edu.rom' image file into sda1 and ran:

c:\> spiflash u edu.rom

...it returned messages indicating success.

Note, what I'm working toward is to make an image of the entire SD drive, so anyone will be able to install by inserting the SD card in another computer, use 'dd' to copy the image file, then put the SD card back into the Gecko.

Gecko Flash BIOS saga

Take 1: DOSBox
I had the idea of booting Quirky from external USB stick, install DOSBox MSDOS emulator, then run the 'spiflash.exe' utility. I realised it is a long shot, as the emulator is just that, an emulated environment.

The experiment itself was pretty straightforward. I booted Quirky, with SDL, SDL_net and dosbox PET packages on the stick, then installed them. The 'spiflash.exe' and 'edu.com' are available on /mnt/sda1...

# dosbox
Z:\> mount c /mnt/sda1
Z:\> c:
Z:\> spiflash u edu.rom
Unknown CPU type

...that is, spiflash executed, put up some preliminary text, then that message about CPU type.

Take 2: MSDOS format
I have reformatted the Gecko internal SD card with sda1 as fat (256MB), sda2 as ext3 (6GB+) and a bit left over in case I want to create a swap partition. I did this with Gparted.

The XP recovery console has 'format' utility, but it does not accept the '/s' (install system shell). So, I'm going to take out the SD card, plug it into another computer that runs Windows, then see if I can run 'format d: /s' (or whatever).

Then I will put the SD card back into the Gecko and should be able to boot to a MSDOS prompt.

...well, that's the theory anyway!

Take 3: Bootable MSDOS CD
Daftdog and bigpup posted some links to boot CDs, I might try them first. Whatever method I come up with, it needs to be easily reproducible by others, so "Take 2" really should be last.

Gecko complaints

I'm typing this message from one of the Geckos right now. This one has WattOS installed, which is based on Ubuntu.

Everywhere I'm looking, I'm finding rough edges, things where I think "they could have done that better".

WattOS
Booting WattOS, some "fail" messages come up, which not what users should be seeing.

After bootup, the touchpad is far too sensitive, and somehow my mouse pointer became translucent with a white translucent rectangle next to it -- I have no idea how that happened, and I only got rid of it by rebooting. I'm now using a USB mouse.

WattOS is quite nice, based on Ubuntu 9.04. The main problem though is the sluggishness compared with Puppy.

Xorg is using the 'rdc' driver, that's good.

When I started Firefox, Shiretoko came up with some kind of error message. can't recall what. Second time, it started without complaint.

I'm finding that I very easily get repeat characters on this keyboard. Hmmm, can't make it happen now.

I need to check that it powers-off. Will do that now.

The Gecko's have arrived

If you haven't been following discussion about the Gecko Edubook on my blog, here are recent blog posts:

http://bkhome.org/blog/?viewDetailed=00771
http://bkhome.org/blog/?viewDetailed=01341
http://bkhome.org/blog/?viewDetailed=01346
http://bkhome.org/blog/?viewDetailed=01353

I picked them up this afternoon, and haven't had time to do much yet. I have fired up one of them, it has Puppy 431 pre-installed. Hmmm, quite a lot wrong though. It looks like they just grabbed the standard 431, knowing that I would be messing around with it and getting everything to work.

Okay, starting from the beginning:

When I opened the packaging, I saw that the metal back-plate was not properly attached -- which someone else also reported. The power cord does not have an Australian plug, but I have an international adapter -- that's something they will need to consider when shipping these things all over the world.

I plugged in the power cord, pressed the power button, and Puppy booted. Snappy performance. But, I quickly noticed these things:

1. Does not power off.
2. "save" button on desktop even though PUPMODE=2 (full "hard drive" install).
3. X is Xvesa.
4. Pupscan: when click to show PCI interfaces, information is broken (that is a problem in 431, it works in Quirky 006).
5. No battery status applet.
6. No audio.
7. No wifi.

I booted Quirky 006 from a USB stick, and noticed a "general protection fault" related to APM -- which would explain why the battery status applet doesn't work. Also, the kernel reports "no acpi support in BIOS found, acpi daemon is unable to proceed".

Running Quirky, Xorg is using the 'vesa' driver.

My Gecko seems to have a different USB wifi unit than what 'pskin' got. It identifies itself as "Realtek RTL8188S wlan adapter", vendor: 0bda, product: 8171. A CD was provided that has a driver for it.

So, I have the kernel patch for sound that pskin posted, source for wifi driver, that leaves video...

Hmmm, will have to investigate the apm/acpi problem too.

A nice little fun project!

Wireless mystery with 2.6.31 kernel

My Acer 3681WXMi laptop has inbuilt wireless that uses the ath5k kernel module. It used to work, however with Quirky and the 2.6.31.5 kernel it doesn't.

In the Network Wizard, when I choose to scan for neworks, an error message comes up that 'ifconfig wlan0 up' failed, with this error:

ifconfig: SIOCSIFFLAGS: Unknown error 132

I googled around, and there are many wireless problems with recent kernels. My particular problem seems to have something to do with the 'rfkill.ko' kernel module, which is now a dependency of the wireless kernel modules.

There is a utility called 'rfkill' that queries the state of a wireless interface. I ran it:

# rfkill list
0: phy0: Wireless LAN
Soft blocked: no
Hard blocked: no

However, after I had done the scan with the Network Wizard, it returns this:

# rfkill list
0: phy0: Wireless LAN
Soft blocked: no
Hard blocked: yes

"Hard blocked" means that there is a hardware switch on my laptop that has turned off the wireless interface. That is just plain incorrect.

I found a "fix" posted in a forum thread discussing the same problem with the 'ath9k' driver. I applied it to my ath5k:

# rmmod ath5k
# rfkill block wifi
# rfkill unblock wifi
# modprobe ath5k
# ifconfig wlan0 up

Somehow, the above clears the problem. But then, running Network Wizard and the problem comes back. So, I had a little play with /usr/sbin/wag-profiles.sh (part of Network Wizard) and modified the cleanUpInterface function:

cleanUpInterface(){

	# put interface down

	#ifconfig "$1" down

	killDhcpcd "$1"

	# kill wpa_supplicant

	killWpaSupplicant "$1"

	# clean up some wireless stuff (taken from wifi-radar)

	if [ "$IS_WIRELESS" = "yes" ] ; then

	  iwconfig "$1" essid off

	  iwconfig "$1" key off

	  iwconfig "$1" mode managed # auto doesn't exist anymore??

	  iwconfig "$1" channel auto

	fi

	# put interface down

	ifconfig "$1" down

	# reset ip address (set a false one)

	ifconfig "$1" 0.0.0.0

	# stop dhcpcd 

	#killall dhcpcd 2>/dev/null

	# flush routing table

	#ip route flush dev "$1"

	# bring interface up again

	#ifconfig "$1" up

	

	#100202

	ERROR="`ifconfig "$1" up 2>&1`"

	if [ "$ERROR" ];then

	 case $1 in

	  wlan*)

	   if [ "`echo "$ERROR" | grep 'ifconfig: SIOCSIFFLAGS: Unknown error 132'`" != "" ];then

	    rmmod $INTMODULE

	    rfkill block wifi

	    rfkill unblock wifi

	    modprobe $INTMODULE

	    ERROR="`ifconfig "$1" up 2>&1`"

	   fi

	  ;;

	 esac

	fi

	if [ "$ERROR" ];then

	  giveErrorDialog "${L_MESSAGE_Failed_To_Raise_p1}${1}${L_MESSAGE_Failed_To_Raise_p2} ifconfig $1 up

$L_MESSAGE_Failed_To_Raise_p3

$ERROR

"

	  return 1

	fi



	return $?

} # end cleanUpInterface

Hmmm, now there is no error message, but it reports no networks found (there is one, about 1 foot away from my laptop!).

I'm far from an expert on networking, and I have left development of these wizards to others. I have messed around a bit, but that's enough.

Now plugging in my USB wireless, which does work with the Wizard.

The Arch Linux guys have been discussing this problem, ath5k/ath9k and 2.6.31.x:

http://bbs.archlinux.org/viewtopic.php?id=81930

...seem to be quite a mess.

Hmmm, I wonder if the 2.6.33 kernel has improved this situation? I wonder if I can/should disable 'rfkill' in the kernel pre-compile configuration?

Rfkill utility links
http://linuxwireless.org/en/users/Documentation/rfkill
http://wireless.kernel.org/download/rfkill/

Termite infestation

This morning I was busy stripping off the inside lining of a wall of my house. Termites!

The little buggers have been there for quite a while. Would you believe it, they eat plasterboard! The wall looked ok, but my finger went straight through. I don't think they eat the plaster itself, rather the fibre in the plasterboard.

My house is wood frame, weatherboard outer cladding, plasterboard and wood panels on the inside. The wood is mostly Jarrah, but there is some pine. The pine is later renovations, not in the original house construction.

Termites (white ants) don't like Jarrah, but they do eat it, slowly. They will look around for something better, and in my case it is the pine and the plasterboard. Interesting, there are some MDF moldings that they ate all around but did not chew at all.

I have not called a pest exterminator, as I don't want to use chemicals. I don't like the whole philosophy of poisoning them. White ants perform an essential function in nature, especially where I live -- there are no worms in this arid soil, but a lot of white ants.

After stripping the wall, I found where they are coming up, and they sure are clever. My house is on stumps, with metal plates separating house and stump. They built an earth "bridge" right around the protruding metal plate. The stump is on the edge of the house, but they built their earth "bridge" on the inside, and I only saw it after getting right down and looking underneath.

Now, my understanding of white ants is that they need access to their nest in the ground. They also need access to underground water. So, if I replace the stump, that breaks their path to home, but my question is, what do those left behind in my house do?

In mean, will they survive? I have stripped the wall, but they have made it up to the roof (not too far fortunately) and there are going to be quite a few that I won't be able to physically remove.

I have remembered an old bush method, to paint the timber with sump oil, the older and dirtier the better. However this site says sump oil doesn't work, but sump oil mixed with creosote does:

http://www.woodworkforums.com/f85/drywood-termites-treated-ply-10105/

...they mention "flying white ants" -- I've had those coming into the house, but no apparent infestation from them -- perhaps they are not the same thing as those in Queensland.

In a day or two I'll tackle rebuilding the stump. There are metal stumps available, that I have seen in Bunnings -- the base gets set in concrete I think.

This page is a good read:

http://www.greenpest.com.au/newsletter_february2009.html

dir2pet, petspec scripts improved

I have improved these two scripts, to help developers with creating the most appropriate entries in each of the package database fields.

These two scripts are in Woof, to be uploaded soon, but as they are of value to anyone using Puppy 4.3.1 or later, I have posted them on the forum:

http://murga-linux.com/puppy/viewtopic.php?t=52033

JWM 464 with i18n patch

I have been adopting shinobar's scripts that he developed for Japanese Puppy, with a view to making Woof (the "Puppy builder") more international-friendly.

Recently, shinobar has patched JWM, the window manager used in most puppies, for i18n support of the menu. Forum post:

http://www.murga-linux.com/puppy/viewtopic.php?t=51847

I have compiled this in Quirky:

http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/jwm2-464-patched_i18n-q1.pet

Builders of other puppies could either compile it themselves, or use my PET package.

Note, Quirky has a recent fribidi package and I had to hack the JWM source:

jwm 464-patched-i18n (patched by shinobar)



edit src/jwm.h:

/*BK #     include  replace with...*/

#     include 

#     include 



# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --disable-confirm --enable-fribidi

# make

Woof uploaded, Jan. 31, 2010

Lots of little fixes to scripts: pmount, drive_all, rc.sysinit, init, xdelta_gui. As reported in recent blog posts.

Xdelta GUI fixed for FAT f.s.

It has been reported that the Xdelta GUI has a problem working with files on a FAT filesystem, as the ':' character is not supported.

I have modified the script /usr/sbin/xdelta_gui to accept .delta files in either format:

oldfile:newfile.delta
oldfile___newfile.delta

And the default is that a .delta is generated in the format:

oldfile___newfile.delta

That is, three underscores is the delimiter.

Pburn 3.1.8

I have upgraded to zigbert's latest Pburn optical drive burner application.

Mplayer recompiled with ALSA enabled

Due to a problem I was having with ALSA in Gnome-mplayer, I compiled 'mplayer' to support oss only.

However, I have modified the Gnome-mplayer package to default to using oss, and there doesn't seem any reason why the mplayer executable cannot be compiled to support both oss and alsa.

Pmount, drive_all scripts change

Shinobar modified these scripts, including internationalization, and these are in Woof, and in Quirky 006.

However, there was one of his modifications that I did not apply, to mount iso9660 filesystems with 'iocharset=utf8'.

Shinobar has informed me that it is necessary to correctly mount CD/DVDs that have been created in Windows.

Ok, I have implemented that change in the scripts. It will be in the next upload of Woof.

Fix detect multiple optical drives

I was testing Quirky 006 on one of my PCs that has two optical drives, a CD and a DVD drive. Gnome-mplayer defaults to playing a DVD at /dev/dvd, however I found that /dev/dvd was missing.

/dev/dvd should get set at bootup, by the /etc/rc.d/rc.sysinit script, to point to the actual optical drive. I found that this automatic setting was failing sometimes when there are two (or more) optical drives.

This is an ancient bug. I have fixed it in rc.sysinit, so it will be in the next upload of Woof.

pkeys boot param fix

Quirky 006 testers have reported that the 'pkeys' boot parameter causes an error message at bootup.

I found that the device node /dev/tty0 was missing from the initramfs, which caused 'loadkmap' and 'loadfont' utilities to fail. I also found that tty1 to tty4 were all the same minor device number, so I fixed that too.

This was a Woof bug.

Busybox experiment

I compiled version 1.16.0 and enabled almost everything. The 'busybox' executable is now 628KB, compared with 427KB in recent puppies.

Over the years I have incrementally replaced Busybox applets with the "full" versions, whenever I found the former to be inadequate.

For Quirky 007 I have decided to experiment with rewinding back to using more of the Busybox applets. For the record, these are the packages that I have changed:

Cut down
coreutils_cut, bc_cut

Removed
bzip2, cpio, diffutils, ed, eject, findutils, grep, gzip, hdparm, less, mktemp, net-tools, psmisc, sed, tar, time, unzip

Left alone
dosfstools, dpkg-deb, ifplugd, ifplugd_old, lzma, module-init-tools, ncurses, pciutils, procps, util-linux, wget.

The packages that I have "left alone", that is, left in Quirky, are those that I know or am pretty certain Busybox cannot adequately replace. Note though, I have left the Busybox applets enabled and they can be executed by 'busybox ' for comparison purposes.

I created 'coreutils_cut' PET package which is very cut down, with only three utilities (install, paste, pr).

I know that we are going to find problems with this and will have to bring back some full utilities, but I'm hoping that it won't be many.

Then there's Bash... I know that there are some scripts written specifically for Bash.
The 'bash' executable is 498KB. Should I be a devil and take that out too? Create a script /bin/bash that just runs /bin/ash?

...so tempting. But I know that doing all of this, especially the Bash thing, is going to cause trouble with compiling some packages. But then, this is Quirky...

Leeman again

Right now it's 7am and I can already feel the heat. So, I'm taking the day off, going to Leeman -- see my recent post about this nice seaside village:

http://bkhome.org/blog/?viewDetailed=01315

But, first, reading the latest posts on the forum. This post about Busybox from technosaurus has made me excited:

http://murga-linux.com/puppy/viewtopic.php?t=51874

It's great that I still get excited about things Puppy-related

Hah hah, but I have already decided to go to Leeman, so checking out the latest Busybox will have to wait. Tonight...

Woof uploaded, January 28, 2010

These are the commit notes, since the last upload of Woof:

ppm fixes, shinobar localized scripts remasterpup2 (bugfix), pmount, drive_all
xwin: make sure xorg kbd layout same as console
this version of Woof used to build Quirky 006

To learn more about Woof:
http://bkhome.org/woof/

Quirky 0.0.6 uploaded

Brief announcement here:

http://distro.ibiblio.org/pub/linux/distributions/quirky/quirky-006/quirky-006-readme.htm

Grab the ISO and 'zdrv" from here:

http://distro.ibiblio.org/pub/linux/distributions/quirky/quirky-006/

...there are also xdelta files for upgrading from 005.

Read my recent blog posts to find out what I have done since 005. As to what is planned for 007, well, I hope to get stuck into the problems with networking. There are some issues with Xorg on certain hardware and I hope to give that some attention also.

ext4 fixed

Gparted did not work with ext4 in Quirky 005. Fixed.

My XO-1 is getting redeployed

I am not using my OLPC XO-1 and have no plans to do so, so I decided the best thing to do is pass it on to someone who can make good use of it.

First, I asked the guy who gave it to me, forum member mdd (Mike), and he replied that he doesn't want it back and I'm welcome to pass it on.

Forum member sculver, who is a support volunteer for OLPC, gave me contact details of a guy who is a OLPC support volunteer in Australia (NSW), first name James.

I contacted James and he replied that yes he can make use of it, either for testing or redeployment. The latter may involve it being sent to Haiti.

So, I'll post it off today or tomorrow.

Note to anyone interested in Puppy Linux running on the XO-1, see the forum: OLPC developer Mitch has succeeded, and sculver, ttuuxxx and others are assisting with solving further issues.

Pmusic 0.9.9.2

I have upgraded to zigbert's latest. This will be in Quirky 006.

lameSMBxplorer 0.1.7b

I have upgraded to Patriot's latest. This will be in Quirky 006.

Grub4dos 0.4.4v1.5.2

I have upgraded to shinobar's latest version. This will be in Quirky 006.

Xorg and console keyboad layouts

I fixed a bug that has been in Puppy for a very long time, maybe always.

When the Xorg Wizard is run, it looks at the console keyboard layout as specified in /etc/keymap and puts an appropriate value into this line in /etc/X11/xorg.conf:

Option "XkbLayout" "us" #xkeymap0

That's fine, however if the Keyboard Wizard is used to change to a different layout, /etc/keymap gets changed and the console layout is updated, however the layout in xorg.conf is not changed.

I have modified /usr/bin/xwin, which is the script that is used to start X, to check for a mismatch between console and Xorg keyboard layout and if found then fix it.

Note, the same situation can occur if the "pkeys" kernel boot parameter is used to change the current setting in /etc/keymap. That is fixed also.

This fix is in Woof, that I will upload soon.

pmount, drive_all scripts

Shinobar posted upgrades for these scripts to the Quirky 005 feedback thread. They are /usr/sbin/pmount and /usr/local/bin/drive_all.

pmount
There was a report on the 005 feedback thread about the CD tray closing too quickly, that this should have fixed. Also it adds locale support.

drive_all
This is the original of the script that runs when a desktop drive icon is clicked on. Shinobar has made some improvements plus locale support. I had recently modified this script, so had to merge my recent changes with shinobar's script.

005 feedback, planning for 006

Thanks guys for testing Quirky 005 and reporting back. I'm working through the 005 forum feedback thread, fixing things. Of course some of these issues are generic to Woof and so apply to all future puppies built with Woof, not just Quirky-specific.

PPM
A couple of bug-fixes. One was changing the choice of repos. Another was 'sysfiles' listed as a dependency of the 'mesa' package.

Crypto modules
As Quirky 005 is built with a cutdown selection of modules, this needs to be tweaked. Further work to be done here, but for now have added all the "crypto" modules, as some needed one may have been missing.

remasterpup2
Shinobar has updated this script, now at version '100120'. There was feedback of a missing value in one of the windows, that this should fix.
I also put the 'ja' locale file into Woof.

Planning for 006
Apart from continuing to work through the 005 feedback, and upgrading some packages, the multimedia is getting another workover. I plan to build 006 with these changes:

Added:
mplayer_dvd
dbus
dbus-glib
gnome-mplayer
gecko-mediaplayer

Removed:
mplayerplug-in
mplayer
ogle
ogle_gui

Note, I created 'mplayer' and 'mplayer_dvd' packages. They are both Mplayer, but the latter has DVD playing support.

Gecko-mediaplayer is a browser plugin and works with gnome-mplayer. They need Dbus to communicate.

Note however, if we decide that gecko-mediaplayer doesn't quite cut it, we can go back to the mplayerplug-in browser plugin, which uses mplayer directly. ...if we did that, then gnome-mplayer does not really need dbus and in theory we could hack it to remove its dbus dependency.

One good thing about mplayerplug-in, gnome-mplayer and gecko-mediaplayer is that they are all developed by the same guy, Kevin, who is very responsive to bug reports.

I can tell you one good thing about gnome-mplayer -- the menu navigation in my DVDs now works. However, the mouse pointer becomes invisible, which makes selection a bit awkward.
Also, there is a problem with switching between full-screen -- can switch to, but when switch back the playing stops and something goes wrong with sound -- have to click the sound icon to bring it back. These are issues that we can report to Kevin!

Woof uploaded, Jan. 26, 2010

This has the Manage System Services added to the BootManager (see previous blog post).

Today is Australia Day!

I went into town, a small group gathered for a couple of speeches, raising of the flag and we sang "Advance Australia Fair". The whole thing took about 15 minutes!

Australia Day originally commemorated the landing at Botany Bay (Sydney) by Captain Cook. Many Aboriginals think of it as "invasion day".

Manage System Services

The BootManager now manages the running of "system services" at bootup.

The main window of BootManager now looks like this:

The Manage System Services window:

The 'HELP' button brings up this:

System services

---------------



"daemon"

...this word is used below. It means an application that when executed, runs

continually in the background.



The scripts in /etc/init.d are executed at bootup and shutdown to start and stop

services.



At bootup, the /etc/rc.d/rc.services script will run all executable scripts

found in /etc/init.d, with the commandline parameter 'start'.

At shutdown, the /etc/rc.d/rc.shutdown script will run all executable scripts

found in /etc/init.d, with the commandline parameter 'stop'.



Puppies built from the Woof build system after January 26, 2010, have System

Service Management provided in the BootManager (see Sysytem menu).

This Services Manager controls which of these scripts will run by setting or

clearing their 'executable' flag -- a script flagged as executable will run,

otherwise not.



System Services Management

--------------------------



Most users do not normally need to disable any of the system services, however

sometimes there might be a need. 



Each service uses CPU and memory resources, so with a slow CPU there may be

some noticeable gain in not running services that are not needed.



On rare occasions a service may cause trouble, so needs to be disabled.



Here are some notes on particular services:



cups

----

This runs the CUPS daemon 'dbusd', required for printing. Leave this enabled unless

you don't need to print.



messagebus

----------

Runs the daemon 'dbusd'.

DBUS is a method for applications to communicate with one another. Only certain

applications use this, and most puppies are built with apps that don't.

If 'messagebus' script is present, it probably means some application is installed

that needs DBUS. An example is 'gecko-mediaplayer' (browser plugin) that uses DBUS

to communicate with 'gnome-mplayer' (multimedia player). So, unless you know

that no apps require this, leave it enabled.



rc.acpi

-------

Runs the daemon 'acpid'.

This is a daemon that provides certain ACPI management functions. Puppy will

still work without it.



rc.firewall

-----------

In most puppies this scipt is actually located at /etc/rc.d. If you have Internet

access then this is essential to provide protection. It doesn't actually launch

any daemon, only loads kernel modules, so runtime resource usage is low. 

The only time that you might want to disable it is when testing the Internet

connection.



slmodem

-------

Runs the daemon 'slmodemd'.

This provides support for some analog dialup modems. There are reports that

this can conflict with sound on some PCs, so if you don't use an analog modem

for Internet access, or a different modem driver, consider disabling this.



sys_logger

----------

This runs the daemons 'syslogd' and 'klogd', which log kernel and application

events (espcially error messages) to various log files, mostly to 

/var/log/messages. This can be disabled and Puppy will still work.



udev

----

Runs the daemons 'udevd' and 'pup_event_frontend_d'.

This is a mechanism that receives information about hardware events from the kernel,

such as a USB pen drive being plugged in or removed. If you don't want automatic

detection of hardware changes while Puppy is running, Puppy will still work

and you will save quite a lot of CPU usage and resources -- worth considering

this one on a very slow CPU.

This one is different from those listed above, as 'udevd' is essential during

bootup. However, it can be killed after bootup -- it involves the daemon 'udevd'

and the daemon 'pup_event_frontend_d' and if disabled these two are killed when

X is started. The technical description is that when X starts, /root/.xinitrc

runs, which launches /sbin/pup_event_frontend_d -- look in that latter script

and you will see that it reads /etc/eventmanager which has a variable 'BACKENDON'

that can be set to kill udevd and pup_event_frontend_d.

There is a GUI manager for this, the EventManager (see System menu), and changing

the 'udev' checkbox will cause the EventManager to run.





Technical notes

---------------



At startup, the system services are executed by /etc/rc.d/rc.services, which

in turn is called from /etc/rc.d/rc.sysinit.



At shutdown, the system servcies are executed (with the 'stop' parameter) by

/etc/rc.d/rc.shutdown.



Puppy does not have runlevels (basically because Busybox doesn't, at least that

was the original reason). Normal Linux distros would have a list of services to

start for each runlevel, but apart from not having runlevels Puppy also only

runs a very minimum essential set of services, that most users would not want

to tamper with.



Note that /etc/rc.d/init.d is a symlink to /etc/init.d



Note, the scripts in /etc/init.d can have any name, but must have their executable

flag set. Any file that does not have the 'x' flag set will be ignored.



Event Management, which is mostly concerned with hardware-related detection and

configuration while X is running, is described in more detail on this web page:



http://puppylinux.com/technical/event-management.htm



Regards,

Barry Kauler

Jan. 2010

The script is /usr/sbin/bootmanager. I also modified /etc/rc.d/rc.sysinit -- syslogd and klogd are no longer started in that script, instead in /etc/init.d/00sys_logger.

Management of "udev" is more involved. The 'udevd' daemon must run at startup, but can be killed later. If the 'udev' checkbox is changed, the Puppy EventManager is launched, which provides the GUI for enablng and disabling udev.

The Manage System Services window can be brought up directly from the commandline by this:

# bootmanager sysdaemons

This is implemented in Woof, that I will upload soon.

US Money Order problem

Today I received a donation of US 20 Dollars, in the form of a US Postal Money Order.

Unfortunately, I cannot cash this. The postmistress at Perenjori made a telephone enquiry to find out if there is some way to cash it, but there isn't.

Hmmm, the fact that it has "negotiable only in the U.S. and possessions" printed on it is a hint!

The person who sent this has only identified themselves as:

Puppy Linux User
Hawthorne
California, USA

...if you are that person, please contact me so that I can return the money order. Or, if you think that you know who that person is, please pass on this message.

Gecko review

I have posted to my blog recently about the Gecko Edubook:

http://bkhome.org/blog/?viewDetailed=00771
http://bkhome.org/blog/?viewDetailed=01341
http://bkhome.org/blog/?viewDetailed=01346

Now, forum member judland (David) has posted a review:

http://www.ipernity.com/blog/virtualsky/223038

Here are some performance criticisms from the review:

About the only thing the Edubook falls short on is in its multi-media / video playback capabilities. Although YouTube videos an MP4s do play, they are very choppy and out of sync with the audio track.

However, there is one saving grace; Puppy Linux come installed with a YouTube video downloader and format converter. If you download the YouTube video you wish to watch and convert it to a lower resolution MPEG, the video plays just fine. Same with MP4s you may have ripped from you DVDs.

and...

Now for the biggest beef I have against the Edubook... the mouse-pad. In my opinion, this mouse-pad is far too sensitive to the touch. I don't even need to be touching the mouse-pad for the mouse pointer to start jumping all over the screen. If I leave a finger resting on one of the mouse buttons, the pad picks it up. If, while typing, my palm comes too close to the surface of the mouse-pad, it registers as a mouse-click. Very annoying.

I wonder if things can be tweaked so that videos can be made to play smoothly when streamed from youtube.com?

It is most likely possible to fix that mouse pad sensitivity problem.

...I look forward to working on these issues when I get my Gecko! And of course I'll post my own mini-review.

There is a comment about a signature needed for credit card payment, yes I had to do that too. It's a bank in Thailand that requires that. I think the alternative is a Western Union transfer.

Mplayerplug-in fixed

As reported recently, The Mplayer browser plugin used in Quirky 005 crashes with some .ram files. After discussion with the developer Kevin, we determined that it is due to the .ram file being a single line of text without a trailing carriage-return character. But kevin pointed out that his code does test for that condition.

Rather than hand it over to Kevin to fix, I decided to tackle it. I found the appropriate file, and where it was crashing. I fixed it.

I have posted the fix to the Mplayerplug-in maillist, also have uploaded the patched source here:

http://bkhome.org/sources/alphabetical/m/mplayerplug-in-20100117-patched.tar.gz

Universal source code highlighter

Forum member droope sent me a pm about this forum thread posted by Dingo:

http://www.murga-linux.com/puppy/viewtopic.php?t=51674

The product is called Highlight and I found the home page:

http://www.andre-simon.de/

Highlight supports these languages:

http://www.andre-simon.de/doku/highlight/en/highlight_langs.html

...Vala, but not Genie.

Droope was prompted to send me the pm as I had once requested such an export feature be added to Nicoedit (color highlighting exported to HTML) -- but that never happened.

I really want Genie, so still waiting. I see in the news page for Highlight that two languages have been added recently, so the author would probably be responsive to a request for Genie.

Flashblock for SeaMonkey

I have been playing with Quirky 005, thinking how to enhance the multimedia control. Recently when I used dialup I was reminded how slow some pages load, so I have installed Flashblock as one more thing to control page content.

I have made it into a PET package and intend that it will be in the next build of Quirky.

SeaMonkey in Quirky 005 already has these addons:

adblock
user-agent-switcher
zombie-keys

Adblock is off by default, as probably most people will find it annoying, however I have left Flashblock on by default. Flashblock can be disabled on a per-page, per-site or totally basis.

Flashblock home:

http://flashblock.mozdev.org/

Woof uploaded, January 20, 2010

This is the Woof version that has been used to build Quirky 005.

I have also updated the Quirky PET packages at ibiblio.org.

Further information on Woof:

http://bkhome.org/woof/

Quirky 0.0.5 uploaded

Brief intro to "multimedia special":

http://distro.ibiblio.org/pub/linux/distributions/quirky/quirky-005/quirky-005-readme.htm

Download from here:

http://distro.ibiblio.org/pub/linux/distributions/quirky/quirky-005/

The ISO is 92.4MB.

I have put in pizzasgood's latest Retrovol tray volume control and sound mixer.

Note that the boot prompt has reverted to "puppy" prefix instead of "quirky", for example "puppy pfix=ram".

Cpu-scaling-ondemand

Forum member trio has updated to version 1.3-2:

http://murga-linux.com/puppy/viewtopic.php?t=45143

Quirky 005 will have this.

Gecko orders

I have just placed an order for two Gecko Edubooks. The manager Michael Barnes has offered one for free, but I am buying a second unit also.

I won't be keeping the second, I'll be donating it to a charity organization in India. A member of an NGO in India, someone I know, will be visiting Perth soon, and I'll give the laptop to her. Hopefully it will arrive in time.

So, it's kind of like the OLPC scheme where you buy one, and one goes to a third world country

Quirky "multimedia special" coming

I have built a Quirky that "plays everything", which will be uploaded as 005, real soon.

I explored radical options in Quirky 003, using Swfdec as the Flash browser plugin, 'ffplay' as the media player, mozplugger to launch appropriate players for embedded browser audio or video, and Ogle to play DVD movies.

Ffmpeg
Yesterday I started again, from scratch. I got ffmpeg out of version-control, dated 2010-01-16. I compiled dependencies opencore-amr 0.1.2 and x264 20091230.

I then went the usual route, compiled xine-lib and gxine, but it was disappointing, some test files did not play properly.

Mplayer
So, I got mplayer out of version-control, dated 2010-01-17, and went through a very tortured process of getting it to compile, with the external ffmpeg dependency. I patched the source, and I will upload 'mplayer-2010-01-17-patched.tar.gz' to my source repo soon -- it also has a special script to do the final link step that builds the 'mplayer' executable. I have also left the source pre-configured so you just type 'make' to compile.

Mplayerplug-in
I compiled mplayerplug-in 20100117 to handle browser embedded content.

Adobe Flash Player
To top it all off, I put in the official Adobe Flash browser plugin, version 9.0.124.0 (more stable than the 9.0.48.0 used in Puppy 431).

Ogle
I built mplayer without DVD playing capability, as I have never been able to get that working right (mplayer can still play VCDs). I have left in Ogle specifically to fill that gap.

Conclusion
The end result is quite remarkable. This combo plays everything I have thrown at it, and plays everything properly. I've got a Lindows RealMedia video that played with garbled sound, now plays nicely. The 'sunrise.wmv' that I had trouble with gaps in the sound, now no gaps.

Beautiful! the iso is 91MB.

Note, one little hiccup is that with some files, mplayer displays an error window that a certain codec is missing, yet the video plays ok. What is happening is that mplayer is thinking a certain codec is required, however ffmpeg has the video and audio capability to play it. I would like to figure out how to suppress that error message window.

Little fix for CD remaster script

Forum member npierce sent me a pm about a problem with the CD remaster script. There is a problem with the 'globicons' file, if the use has made any changes to the desktop icons. This file is stored in two places, and the one that has preference does not get saved by the remaster script. Fixed. I also added saving of the rox 'Options' file.

The script is /usr/sbin/remasterpup2.

Audio/video test site

Forum member coolpup often pm's me with useful information, such as notification of a new package version or a useful link. Like this one:

http://www.yolinux.com/TUTORIALS/LinuxTutorialMimeTypesAndApplications.html

That's very good. I'm just about to get back onto developing the multimedia in Quirky and of course want to do lots of testing.

One-file Quirky for OLPC XO-1

Well, an attempt anyway. Here it is:

http://distro.ibiblio.org/pub/linux/distributions/quirky/test/vmlinuz-16JAN2010-GEODE-OLPC.t01

This is a 'vmlinuz' Linux kernel with everything embedded in it. It is 81.74MB.

Xcore86 CPU in Gecko Edubook

I have been tracking down technical details on the Xcore86 CPU used the the Gecko Edubook laptops. The website for the Xcore86:

http://www.xcore86.com/

However that site is severely lacking in technical detail. However, I discovered from this wikipedia page that the Xcore86 is a rebadged Vortex86:

http://en.wikipedia.org/wiki/Vortex86

The Gecko uses a Vortex86MX chip, and there is a PDF technical specifications file:

ftp://download@ftp.dmp.com.tw/vortex86dx/Vortex86DX_V0.9A_Brief.pdf

...well, that is for the "DX" model, which apparently has the same CPU core. Um, need to do better than that, ok, the pmx-1000 is the consumer-grade version of the Vortex86MX and here is a technical specification file:

http://www.vortex86mx.com/wp-content/uploads/2009/03/PMX-1000_V10_Brief.pdf

...no, that one is very brief, hardly any information.

Anyway, what I can glean from the "DX" specs, the CPU core is a equivalent to a 486SX CPU, plus FPU (maths processor) and GPU (graphics processor) plus lots of other things are on the chip.

The chip is actually manufactured by DMP Electronics in Taiwan:

http://dmp.com.tw/
http://www.vortex86mx.com/

The 486 CPU does not have the MMX instructions, nor the MTRR registers. The kernels that we compile for Puppy are for the "i486", which basically means "486sx + FPU". One thing that is enabled in the kernel config is support for MTRRs, but I have done some research on this and apparently it doesn't matter, the kernel will still work on a CPU that does not have the MTRR registers.

There is a question mark over Xorg and whether it needs the MTRRs. I'll have to check, perhaps that is also configurable prior to compiling the Xorg servers.

There may be some multimedia apps used in Puppy that are configured to expect the MMX instructions, or even some more advanced architectures such as SSE and SSE2. It will be necessary to check when configuring those apps prior to compiling that if these features are enabled, the app can still work if they aren't there.

Anyway, overall conclusion is that the CPU used in the Gecko Edubook is Puppy-compatible, but we will need to be mindful that it lacks MTRR, MMX, SSE, SSE2 when configuring some applications prior to compiling.

Now to confuse the issue... quoting from this site:
http://hkampar.moh.gov.my/modules/news/article.php?storyid=31

The Xcore86 (also known as the PMX 1000) is x586 based System on Chip (SoC) that offers a less than average thermal envelope of the Atom. The Xcore86 uses a x586 core, MMX compatible extensions, Video Core and r6040 Lan on Chip from RDC, HD sound core from Realtek and electronic integration from DMP.

...huh, "MMX compatible extensions"? Why isn't that in any of the official docs?

Creating one-file Quirky for XO-1

I looked through the kernel .config file that mavrothal posted on the forum. I didn't follow it exactly, so there is some element of gamble here, but anyway I have now created a one-file Quirky with kernel configured for the Geode LX CPU and (hopefully) the OLPC XO-1.

I'm building it now and will announce when it is uploaded.

For the XO-1.5, I would need to get a kernel .config file (ideally for the 2.6.31.x kernel) to see what differences there are. probably there are some hardware differences, apart from the CPU.

Another 'vmlinuz' for OLPC test

I recompiled the one-file Quirky, this time I configured the kernel for the Geode GX1 CPU.

Download here:
http://distro.ibiblio.org/pub/linux/distributions/quirky/test/vmlinuz-15JAN2010.t00

Note, I also turned off MTRR register support in the kernel, as I don't know if the Geode GX1 has those registers. I think that Xorg needs those registers.
Anyway, see if this boots, and I can do another compile with MTRR enabled.

I had a quick look at Geode specs, I see that the GX1 doesn't even have the MMX instructions, but the GX2 does. If I recall rightly, even the i486 has MMX instructions.
I think that some applications compiled for a i486 are not going to work with a Geode GX1.

I don't know whether the first XO's have a GX1 or GX2, hopefully not the former.

Notes on booting Quirky on OLPC

Continuation from previous blog post, about the single-file Quirky:

http://bkhome.org/blog/?viewDetailed=01337

My OLPC laptop doesn't work. It has been sitting in its original box, with original packing, since 2008, unused. This is my blog report back then when I tested it:

http://www.puppylinux.com/news/news400a5-400a7.htm

The battery would be discharged after such a long time. I have the power adaptor connected, I get the orange battery light. If I press the power button, the green power-on lights up, but the screen remains black. I thought maybe I needed to wait awhile for the battery to charge, but after one hour, still no-go.

Anyway, I have been reading-up on booting. A good reference:

http://wiki.laptop.org/go/Olpc.fth

...I don't know how much of that applies to my early-model XO.

Anyway, it seems pretty straightforward. If you have a normal bootable USB Flash drive, with one partition, say 1GB drive, fat16 filesystem. Perhaps play safe and use a small Flash drive of 1GB and fat16.

On the Flash drive, place my Quirky file, 'quir004.t00', rename it to 'vmlinuz' -- no particular reason, just playing safe, use a "normal" name for the kernel.

Create a file /boot/olpc.fth on the Flash drive, with this in it:

unfreeze
boot u:\vmlinuz root=sda1 pmedia=usbflash

...docs indicate the "root=sda1" is required, it would be interesting to see if it worked without that parameter.

If booting from a SD card:

unfreeze
boot sd:\vmlinuz root=mmcblk0p1

The above is theoretical, as I can't test it myself!

Clock battery
Read this about the clock battery in the OLPC. Maybe mine is dead.
http://en.forum.laptop.org/viewtopic.php?f=1324&t=164934

Test single-file Quirky

As I have previously announced, also on the Quirky home page, Woof can build a distro which is just one file, the Linux kernel, with absolutely everything inside it to boot to the desktop. No access to any drives required -- if you can get the kernel to start booting, that's it, you should get a desktop.

The single-file Quirky has some issues that I still need to sort out, but Raffy and others involved in getting Puppy to boot on the OLPC have asked about trying this. So here it is.

Readme file:
http://distro.ibiblio.org/pub/linux/distributions/quirky/test/quirky004-readme.htm

Download live-CD: URL CORRECTION:
http://distro.ibiblio.org/pub/linux/distributions/quirky/test/quirky-004-t00.iso
...you can easily mount this and extract the kernel file.

Md5sum:
http://distro.ibiblio.org/pub/linux/distributions/quirky/test/md5sums.txt

Quirky home page:
http://bkhome.org/quirky/

The ISO is 92MB, a bit big as I have included all the modules.

Note that any Puppy built with Woof can be built this way, it is not just a Quirky thing. It does require an appropriate .config file for the kernel source, and currently only the 2.6.31.5 kernel PET package has that (it's on the ibiblio.org Quirky repository).

URGENT:
I uploaded the wrong iso. Sorry, it was late, I was tired and my attention lapsed. Uploading the correct iso now. Will post when it's done.

CORRECTION:
The correct iso has now been uploaded, see above. Note, the ftp connection broke halfway through uploading, and I had to resume, so check the md5sum.

Woof uploaded, January 13, 2010 (again)

Full hd install fixed!

Testers of Quirky 003 and earlier have reported that a full hard drive installation is broken. In fact, building any puppy, Dpup, Upup, or whatever, the full hd install is broken.

The reason for this is recent optimizations that I made in Woof, in which files in the initrd are not duplicated in the main f.s.:

http://bkhome.org/blog/?viewDetailed=01294
http://bkhome.org/blog/?viewDetailed=01218

...full hd install would have been partly broken right back in late November 2009.

So, as there are lots of people who like doing a full hd install, I have immediately uploaded the fix. For information about Woof and how to download, see here:

http://bkhome.org/woof/

Note, the script that I fixed is /usr/sbin/puppyinstaller.

Nouveau merging in 2.6.33 kernel

I tested the nouveau Xorg driver on my PC with Nvidia 6600 GT card, but no-go. I reported recently on how I compiled it:

http://bkhome.org/blog/?viewDetailed=01332

The good news is that the nouveau kernel driver will be merging in the 2.6.33 kernel, which hopefully should greatly improve the nouveau Xorg driver. The news about this is here:

http://lwn.net/Articles/366648/

...I wouldn't hold my breath though, I don't think the Xorg nouveau driver is likely to suddenly become really great.

Some useful links for developers:

http://nouveau.freedesktop.org/wiki/InstallNouveau
http://nouveau.freedesktop.org/wiki/InstallDRM
http://people.freedesktop.org/~pq/nouveau-drm/

Woof uploaded, January 13, 2010

Commits file comments:

ppm fix install t2 .tar.bz2 pkgs, fix xorgwizard when choose generic vesa, quirky: xorg-server, xf86-video-nv upgrades
xorgwizard: alternate driver selection, libdrm,nv,nouveau: xorg drivers in quirky

Details in recent blog posts.

Xorg Wizard enhanced

I have made a significant improvement to the Xorg Wizard, script /usr/sbin/xorgwizard.

Often there are alternative drivers to choose from, for example:

intel i810 vesa
nv nouveau vesa
ati radeon vesa

I have setup a mechanism for easy switching between these alternatives. Directory /usr/lib/xorg/modules/drivers-alternate is where unused drivers are kept. I have put the old i810 driver (from Xorg 7.2) and the nouveau driver in there.

For Intel video hardware, you can test the intel and/or vesa drivers, but there is now an option to swap-in the i810 and swap-out the intel driver, then re-probe.

Ditto for nv and nouveau. They can be swapped, so that when Xorg probes it only sees one of them.

It was very easy on my laptop, with Intel hardware, to test each of the vesa, intel and i810 drivers. I have not yet built a Quirky CD and tried it on my daughter's old PC (which has an Nvidia 6600 GT card).

I'll upload Woof soon with this improvement.

Compiling Xorg Nouveau driver

This has turned up some interesting stuff. I got xf86-video-nouveau from git, dated 2010-01-12.

util-macros
I had trouble configuring, then found that this package is required. I have installed version 1.4.1. I'll make sure that this is in the Quirky 'devx'.

aclocal
When I ran 'autogen.sh', to generate the 'configure' script, there was an error:

./configure: line 10956: syntax error near unexpected token `RANDR,'
./configure: line 10956: `XORG_DRIVER_CHECK_EXT(RANDR, randrproto)'

I found the reason here:

http://nouveau.freedesktop.org/wiki/FAQ#head-6f41727cc579ed3d7f42d67328466e2acbda9fc4

I created a file /usr/share/aclocal/dirlist, with this in it:

/usr/X11R7/share/aclocal

I have put this fix into the 'rootfs-skeleton' in Woof.

libdrm_nouveau
autogen.sh still gave an error:

checking for LIBDRM_NOUVEAU... configure: error: Package requirements (libdrm_nouveau) were not met:
No package 'libdrm_nouveau' found

...this is supposed to be part of package 'libdrm', that I do have installed. Hmmm. Ok, I am compiling libdrm, version 2.4.17 (the latest, 20 dec 2009):

UPDATE 13Jan2010: need CFLAGS to make it compile libdrm_intel:

# CFLAGS="-march=i486" ./configure --prefix=/usr/X11R7 --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu --enable-nouveau-experimental-api --enable-radeon-experimental-api --with-kernel-source=/usr/src/linux-2.6.31.5 --enable-intel
# make
# new2dir make install

xf86-video-nouveau 2010-01-12
We should be good to go now:

# ./autogen.sh
# ./configure --prefix=/usr/X11R7 --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu
# make

oh crap...
nv_accel_common.c: In function 'NVAccelInit2D_NV50':
nv_accel_common.c:470: error: 'NV50_2D_CLIP_ENABLE' undeclared

...I looked in an older version of xf86-video-nouveau and the actual numeric value is used, which I have substituted. It looks like the developers have some extra header file that I don't have.

But then it stops again, more undefined symbols. These symbols are undefined anywhere, can't find them by googling either:

nv10_exa.c: In function 'setup_combiners':
nv10_exa.c:493: error: 'NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE0' undeclared (first use in this function)
nv10_exa.c:493: error: (Each undeclared identifier is reported only once
nv10_exa.c:493: error: for each function it appears in.)
nv10_exa.c:495: error: 'NV10TCL_RC_IN_RGB_A_MAPPING_UNSIGNED_INVERT' undeclared (first use in this function)
nv10_exa.c:498: error: 'NV10TCL_RC_IN_RGB_B_INPUT_TEXTURE1' undeclared (first use in this function)
nv10_exa.c:500: error: 'NV10TCL_RC_IN_RGB_B_MAPPING_UNSIGNED_INVERT' undeclared (first use in this function)

The file 'nv10_exa.c' has had a lot of work done on it recently. So, it's another case of the developers having the definitions of those symbols, but not me.

Finally...
Oh, man, it took me awhile. I tried several different versions of xf86-video-nouveau. I got some courtesy of the Arch guys:

http://mirror.leaseweb.com/archlinux/other/xf86-video-nouveau/

I tried most of them, got different errors. Finally, I tried 'xf86-video-nouveau-20091221.tar.bz2' and finally, it compiled. Yippee. It has to be the right versions of everything, and I finally got it. My guess is that libdrm is being developed in-step, so you must have the right version of libdrm.

Xorg server, nv, updated

I decided to update the 'xorg-server' and 'xf86-video-nv' packages in Quirky. As Quirky is cut-down, I thought that I would probably need the 'mesa' package, so I ran the Puppy Package Manager and installed it from the T2 binary repository.

Puppy Package Manager fixed for .tar.bz2 packages
PPM also stated that 'mesa' needed two dependencies, 'dri2proto' and 'sysfiles'. I installed the former only. 'sysfiles' is core structural files that are already in Puppy, so to avoid this issue in the future, I added "sysfiles" to the 'PKG_NAME_IGNORE' variable in file /root/.packages/PKGS_MANAGEMENT (this is done in Woof, so will be in all puppy builds).

But then I discovered that PPM has a bug, does not actually install any files. I edited /usr/local/petget/installpkg.sh and fixed so that .tar.bz2 packages do get installed.

I ended up installing these T2 binary pkgs:

mesa
dri2proto
glproto

I have the 'devx' and the 2.6.31.5 kernel source SFS files loaded, although I don't know if the latter is needed.

xorg-xserver 1.7.3
Quirky is supposed to be as cutdown as is reasonable... so I have taken out the XvMC (Xv Motion Control) support, as my understanding is that not all Xorg drivers support it. DRI2 is probably in the same boat, but I have left that in. It is a bit of a concern taking out too much, as then the commercial Nvidia drivers might not compile -- although one of my reasons for upgrading is to see if the latest 'nv' driver will do the job, so the commercial Nvidia drivers won't be needed. Anyway, giving this a go...

# ./configure --prefix=/usr/X11R7 --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu --disable-dmx --disable-xvfb --disable-xephyr --disable-config-hal --disable-xnest --enable-kdrive --enable-xsdl --disable-xvmc
# make
# new2dir make install

...very interesting, I enabled the 'Xsdl' kdrive server. What is this? There is no documentation in the package. Ah, the Xorg 7.5 release notes have this:

Xsdl server
The experimental Xsdl server has never been finished or maintained, and will be removed in future X server releases.

...so much for that! There is not much left in the 'kdrive' category. For 7.5, everything has been removed, except for the Xsdl and Xfbdev servers.

xf86-video-nv 2.1.16
The version compiled in T2 and hence in Quirky up until 003 is 2.1.15. So, we shall see if this one makes any difference...

# ./configure --prefix=/usr/X11R7 --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu --with-xserver-source=../xorg-server-1.7.3
# make
# new2dir make install

These upgrades will be in Quirky 004. The bugfixes for Woof will be uploaded soon.

SDL compile howto

Just for the record, this is how I compiled SDL and the extras for Quirky. I compiled SDL recently as it is needed for the 'ffplay' utility in ffmpeg. Then a couple of days ago I compiled the extras:

SDL 1.2.14

# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu --disable-esd --disable-pulseaudio --disable-arts --disable-mintaudio --disable-video-cocoa --disable-video-directfb --disable-video-opengl --enable-nasm --disable-video-photon --disable-video-carbon --disable-video-ps2gs --disable-video-ps3 --disable-video-ggi --disable-video-vgl

# make

# new2dir make install



sdl_mixer 1.2.11

# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu --disable-music-mod --disable-music-midi --enable-music-mp3-mad-gpl

# make

# new2dir make install



SDL_image 1.2.10

# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu

# make

# new2dir make install



SDL_ttf 2.0.9

# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu 

# make

# new2dir make install

Note that the default of SDL_mixer is to require the 'smpeg' package to provide mp3 audio support. However, the '--enable-music-mp3-mad-gpl' option makes it use 'libmad' instead.

Even so, when I was compiling SDL games, I came across some that require the 'smpeg' package.

Just a little note on the side:
this could be completely wrong, I just got the vibes that things have moved on since SDL was at it's heyday. A bit like Tcl/Tk. Libraries like 'cairo' in GTK have made GTK coding of games more than competitive? I dunno, this is a very superficial assessment, as I say, just some vibes I got while browsing. You are welcome to shoot me down on this one.

Xorg Wizard 'vesa' bugfix

I found a couple of bugs in the Xorg Wizard. After the user has chosen to test the video, a dialog window asks if want to change to the generic vesa driver -- but the button display only instead of the appropriate labels. Furthermore, if the vesa generic driver is chosen, a following window after the next test states that the previous driver is still chosen. I fixed both bugs.

Nvidia GeForce 6600 GT
Sometime ago, my daughter gave me her old computer. Today is the first time I have plugged it in. No hard drive. What interests me about this PC is that it has an Nvidia card, a GeForce 6600 GT model.

Ok, I get video for both the 'nv' and 'vesa' drivers, however the ps/2 mouse is crazy -- pointer jumps one side of the screen to the other, or just moves horizontally along the bottom, or disappears entirely. Totally unusable.

Well, I'll google around, see if others have this problem.

More SDL games

A few more:

Sizes of PETS:
Burgerspace: 130KB, Cosmosmash: 70KB, Lmarbles: 764KB, Wakkabox: 76KB.

http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/burgerspace-1.8.3-q1.pet
http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/cosmosmash-1.4.3-q1.pet
http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/lmarbles-1.0.8-q1.pet
http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/wakkabox-1.1.0-q1.pet

That's enough, I've run out of steam!

SDL games

Stardust 004

So, what's happening on the Puppy 4.x front? There are a few puplets being developed. I noticed in the forum today two that are especially active: Jemimah's Puppeee (for eeePC netbooks) and zigbert's Stardust.

Stardust has now reached version 004. Read more:

http://murga-linux.com/puppy/viewtopic.php?t=51126

Retrovol 0.1, Xtmix 0.4

Retrovol
Pizzasgood has written a very nice volume control for the tray, with a popup mixer. Pizzasgood has achived a fully-featured mixer, as some other GUI mixers are somewhat lacking -- forcing us to use the ncurses text-mode AlsaMixer.

Forum thread:
http://www.murga-linux.com/puppy/viewtopic.php?t=50744

Pizzasgood, I would like to make a request. Could you add a commandline option to bring up the mixer immediately -- and it would have to run with some other name than "retrovol" so that it doesn't get swallowed by the JWM tray. That would be really neat, then I can have a menu entry for the mixer.

EDIT: no sorry, Retrovol is not a window that gets swallowed, as in the example of Xtmix, so running the mixer as a different name doesn't apply.

Xtmix
Note that Quirky 003 has Xtmix, an oldy-but-goody. In /root/.jwmrc-tray there is this:

xtmix -launch

...this causes a round volume control knob to display, that gets swallowed in the JWM tray (it's running name is "xtmix-launcher". Right-click brings up a mixer. There is also a menu entry, which launches 'xtmix' which brings up the mixer directly.
The Xtmix mixer is a bit light-on as far as slider controls goes, and that's the main disadvantage. What is really good is the tray applet uses few resources.

The Xtmix PET:
http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/xtmix-0.4-q1.pet

And the source:
http://bkhome.org/sources/alphabetical/x/xtmix04.tar.gz

GtkLP print files, manage queue

I have compiled GtkLP 1.2.7 for Quirky:

http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/gtklp-1.2.7-q1.pet

I looked at a GtkLP package created by ttuuxxx:

http://www.murga-linux.com/puppy/viewtopic.php?p=371589&sid=a2bf25bc0fc4b7faa2a5dfaf0acff7f4

ttuuxxx has modified /etc/gtk-2.0/gtkrc, setting this:

gtk-print-backends = "lpr,file"

...whereas the default is "file,cups".

Ttuuxxx also has a script /usr/bin/lpr.

Ttuuxxx's package is for Puppy 431 I think, however in Quirky I don't know of any printing problem that requires such a solution. So, my package doesn't change 'gtk-print-backends' and has the default lpr script that comes with the cups package.

Printing note
Quirky uses CUPS 1.3.11. Interesting, it doesn't ever ask for username/password -- I somehow seem to have fixed that.

Compiling note
Quirky has Xorg in /usr/X11R7, but I have got it setup so never have trouble anymore with configuring and compiling packages. They are all able to find the libs. This contrasts with trouble I had sometime in Puppy4.

g2calc won't compile

I have tried to compile g2calc version 0.13 created by disciple:

# cd src
# make
yacc -d --report=solved --report=itemset expcalc_par.y
conflicts: 44 reduce/reduce
flex expcalc_tok.l
gcc -c -o y.tab.o y.tab.c
In file included from expcalc_par.y:129:
expcalc_tok.l: In function 'yylex':
expcalc_tok.l:53: error: 'tE' undeclared (first use in this function)
expcalc_tok.l:53: error: (Each undeclared identifier is reported only once
expcalc_tok.l:53: error: for each function it appears in.)
make: *** [y.tab.o] Error 1

Hmmm, I downloaded g2calc original source, it compiles, but it doesn't work -- I click buttons, nothing displays. Oh well.

lameSMBxplorer 0.1.1a

Patriot has developed a new Samba shares browser, that looks like it can replace HairyWill's Pnethood. Forum thread:

http://murga-linux.com/puppy/viewtopic.php?t=50901

I have made it into a PET package and placed at my Quirky repository:

http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/lamesmbxplorer-0.1.1a.pet

Of course, Patriot already had a PET package, but I updated it with the 'pet.specs' file and a list of dependencies. Which are:

+gtkdialog3,+nbtscan,+samba_client

One difference from Pnethood is that lameSMBxplorer does not use 'mount.cifs' from the 'mountcifs' package -- or does it?

pet.specs
A note about the database entry in 'pet.specs', this is it for lamesmbxplorer:

lamesmbxplorer-0.1.1a|lamesmbxplorer|0.1.1a||Network|76K|pet_packages-quirky|lamesmbxplorer-0.1.1a.pet|+gtkdialog3,+nbtscan,+samba_client|connect to samba shares|||official|

...I'm planning to make a small change, allow the subdirectory field to be empty. I have put "pet_packages-quirky" in it, but that of course is only relevant when it is at my repo.
I will now allow the field to be blank:

lamesmbxplorer-0.1.1a|lamesmbxplorer|0.1.1a||Network|76K||lamesmbxplorer-0.1.1a.pet|+gtkdialog3,+nbtscan,+samba_client|connect to samba shares|||official|

...and the meaning of the empty field is that the package is located in the default location on whatever repo it is held.

I intend to update the package manager to recognise an empty path field to have that meaning.

PupScan fixed

The PupScan application (see System menu) was displaying incorrect information about the PCI devices.

The problem is that it uses the 'scanpci' utility, which I have in Woof, but it is no longer part of Xorg. It used to be, up to Xorg 7.3, that the libpciaccess library had a database of PCI information, however from 7.4 onwards the libpciaccess does not have its own database, instead reads from /usr/share/pci.ids (which is part of pciutils package).

So Woof builds now have the full 'pci.ids' file, but 'scanpci' is still included. However, it no longer works properly.

So, I have modified /usr/sbin/pupscan so that it detects existence of /usr/share/pci.ids and if exists then uses the lspci utility to obtain the PCI information.

Woof uploaded, January 7, 2010

These are the comments in the 'commits' file since the last upload:

ntfs-3g in initrd (and quirky pkg) updated from 2009.4.4 to 2009.11.14
fix shutdown if ntfs partition mounted, 60-sec timeout in shutdown dlg

JorgenS reported in the Quirky 003 feedback thread that the on-going problem remains, if a NTFS partition is mounted, Quirky will not shutdown, X just restarts. I have isolated the problem, and fixed /etc/rc.d/rc.shutdown.
Note, this problem did not occur if the NTFS partition has the 'pupsave'.

There was another issue where unattended shutdown may be required, and where there is no 'pupsave' file. The rc.shutdown script brings up a dialog asking if you want to create a 'pupsave', but if unattended then it will never shutdown. I have put in a 60 second timeout in the dialog, after which it will shutdown without creating a 'pupsave'.

Hacking Pwireless2

There are some issues with putting Jemimah's Pwireless2 beta2.7 into Quirky.

Jemimah has included binary executables from 'gtrayicon', 'dhcpcd' and 'wpa_supplicant' packages inside Pwireless2. That is not acceptable for me, so I am in the process of providing those packages as separate PETs.

dhcpcd
I recently compiled version 5.1.3. There are problems with using this with Pwireless2. I configured like this:

./configure --libexecdir=/lib/dhcpcd --dbdir=/var/lib/dhcpcd --sysconfdir=/etc

as this is the posix recommended layout, as stated in the dhcpcd docs. However, jemimah has left the db-directory as /var/db, the execdir as /usr/libexec and the binary executables are in /usr/local/sbin whereas I have them in /usr/sbin -- and Jemimah's scripts have hardcoded paths for the excutables (which she probably did to avoid using old executables (?)).

I have created a dhcpcd PET package with symlinks, which fixes most of the problems. However, /var/db is used by something else and I cannot make that into a symlink. Umm, it seems only /root/Startup/Pwireless2 explicitly addresses /var/db, so I request that another line be inserted in the script:

rm -f /var/db/dhcpcd*
rm -f /var/lib/dhcpcd/dhcpcd*

Oh, I have just looked in Jemimah's /etc/dhcpcd.conf, it has this line:

denyinterfaces lo,ppp0,wmaster0

...hmmm, I need to put this on hold until after further discussion with Jemimah and others. Um, wouldn't it be better, if dhcpcd.conf is modified in a manner that may adversely affect other uses (like ppp0 dialup), that the dhcpcd executable be called with "--config "? The dhcpcd commandline also allows "--denyinterfaces ".

wpa_supplicant
Jemimah has configured this differently from the one already in Puppy or Quirky. She has sent me the configure details but I have not yet got onto that. For the record, this is the .config file that she has sent me:

# Example wpa_supplicant build time configuration

#

# This file lists the configuration options that are used when building the

# hostapd binary. All lines starting with # are ignored. Configuration option

# lines must be commented out complete, if they are not to be included, i.e.,

# just setting VARIABLE=n is not disabling that variable.

#

# This file is included in Makefile, so variables like CFLAGS and LIBS can also

# be modified from here. In most cases, these lines should use += in order not

# to override previous values of the variables.





# Uncomment following two lines and fix the paths if you have installed OpenSSL

# or GnuTLS in non-default location

#CFLAGS += -I/usr/local/openssl/include

#LIBS += -L/usr/local/openssl/lib



# Some Red Hat versions seem to include kerberos header files from OpenSSL, but

# the kerberos files are not in the default include path. Following line can be

# used to fix build issues on such systems (krb5.h not found).

#CFLAGS += -I/usr/include/kerberos



# Example configuration for various cross-compilation platforms



#### sveasoft (e.g., for Linksys WRT54G) ######################################

#CC=mipsel-uclibc-gcc

#CC=/opt/brcm/hndtools-mipsel-uclibc/bin/mipsel-uclibc-gcc

#CFLAGS += -Os

#CPPFLAGS += -I../src/include -I../../src/router/openssl/include

#LIBS += -L/opt/brcm/hndtools-mipsel-uclibc-0.9.19/lib -lssl

###############################################################################



#### openwrt (e.g., for Linksys WRT54G) #######################################

#CC=mipsel-uclibc-gcc

#CC=/opt/brcm/hndtools-mipsel-uclibc/bin/mipsel-uclibc-gcc

#CFLAGS += -Os

#CPPFLAGS=-I../src/include -I../openssl-0.9.7d/include \

#	-I../WRT54GS/release/src/include

#LIBS = -lssl

###############################################################################





# Driver interface for Host AP driver

CONFIG_DRIVER_HOSTAP=y



# Driver interface for Agere driver

#CONFIG_DRIVER_HERMES=y

# Change include directories to match with the local setup

#CFLAGS += -I../../hcf -I../../include -I../../include/hcf

#CFLAGS += -I../../include/wireless



# Driver interface for madwifi driver

#CONFIG_DRIVER_MADWIFI=y

# Set include directory to the madwifi source tree

CFLAGS += -I../../madwifi



# Driver interface for Prism54 driver

# (Note: Prism54 is not yet supported, i.e., this will not work as-is and is

# for developers only)

#CONFIG_DRIVER_PRISM54=y



# Driver interface for ndiswrapper

#CONFIG_DRIVER_NDISWRAPPER=y



# Driver interface for Atmel driver

CONFIG_DRIVER_ATMEL=y



# Driver interface for old Broadcom driver

# Please note that the newer Broadcom driver ("hybrid Linux driver") supports

# Linux wireless extensions and does not need (or even work) with the old

# driver wrapper. Use CONFIG_DRIVER_WEXT=y with that driver.

#CONFIG_DRIVER_BROADCOM=y

# Example path for wlioctl.h; change to match your configuration

#CFLAGS += -I/opt/WRT54GS/release/src/include



# Driver interface for Intel ipw2100/2200 driver

CONFIG_DRIVER_IPW=y



# Driver interface for Ralink driver

CONFIG_DRIVER_RALINK=y



# Driver interface for generic Linux wireless extensions

CONFIG_DRIVER_WEXT=y



# Driver interface for Linux drivers using the nl80211 kernel interface

#CONFIG_DRIVER_NL80211=y



# Driver interface for FreeBSD net80211 layer (e.g., Atheros driver)

#CONFIG_DRIVER_BSD=y

#CFLAGS += -I/usr/local/include

#LIBS += -L/usr/local/lib



# Driver interface for Windows NDIS

#CONFIG_DRIVER_NDIS=y

#CFLAGS += -I/usr/include/w32api/ddk

#LIBS += -L/usr/local/lib

# For native build using mingw

#CONFIG_NATIVE_WINDOWS=y

# Additional directories for cross-compilation on Linux host for mingw target

#CFLAGS += -I/opt/mingw/mingw32/include/ddk

#LIBS += -L/opt/mingw/mingw32/lib

#CC=mingw32-gcc

# By default, driver_ndis uses WinPcap for low-level operations. This can be

# replaced with the following option which replaces WinPcap calls with NDISUIO.

# However, this requires that WZC is disabled (net stop wzcsvc) before starting

# wpa_supplicant.

 CONFIG_USE_NDISUIO=y



# Driver interface for development testing

#CONFIG_DRIVER_TEST=y



# Driver interface for wired Ethernet drivers

CONFIG_DRIVER_WIRED=y



# Driver interface for the Broadcom RoboSwitch family

CONFIG_DRIVER_ROBOSWITCH=y



# Enable IEEE 802.1X Supplicant (automatically included if any EAP method is

# included)

CONFIG_IEEE8021X_EAPOL=y



# EAP-MD5

CONFIG_EAP_MD5=y



# EAP-MSCHAPv2

CONFIG_EAP_MSCHAPV2=y



# EAP-TLS

CONFIG_EAP_TLS=y



# EAL-PEAP

CONFIG_EAP_PEAP=y



# EAP-TTLS

CONFIG_EAP_TTLS=y



# EAP-FAST

# Note: Default OpenSSL package does not include support for all the

# functionality needed for EAP-FAST. If EAP-FAST is enabled with OpenSSL,

# the OpenSSL library must be patched (openssl-0.9.8d-tls-extensions.patch)

# to add the needed functions.

#CONFIG_EAP_FAST=y



# EAP-GTC

CONFIG_EAP_GTC=y



# EAP-OTP

CONFIG_EAP_OTP=y



# EAP-SIM (enable CONFIG_PCSC, if EAP-SIM is used)

#CONFIG_EAP_SIM=y



# EAP-PSK (experimental; this is _not_ needed for WPA-PSK)

#CONFIG_EAP_PSK=y



# EAP-PAX

#CONFIG_EAP_PAX=y



# LEAP

CONFIG_EAP_LEAP=y



# EAP-AKA (enable CONFIG_PCSC, if EAP-AKA is used)

#CONFIG_EAP_AKA=y



# EAP-AKA' (enable CONFIG_PCSC, if EAP-AKA' is used).

# This requires CONFIG_EAP_AKA to be enabled, too.

#CONFIG_EAP_AKA_PRIME=y



# Enable USIM simulator (Milenage) for EAP-AKA

#CONFIG_USIM_SIMULATOR=y



# EAP-SAKE

#CONFIG_EAP_SAKE=y



# EAP-GPSK

#CONFIG_EAP_GPSK=y

# Include support for optional SHA256 cipher suite in EAP-GPSK

#CONFIG_EAP_GPSK_SHA256=y



# EAP-TNC and related Trusted Network Connect support (experimental)

#CONFIG_EAP_TNC=y



# Wi-Fi Protected Setup (WPS)

#CONFIG_WPS=y



# EAP-IKEv2

#CONFIG_EAP_IKEV2=y



# PKCS#12 (PFX) support (used to read private key and certificate file from

# a file that usually has extension .p12 or .pfx)

CONFIG_PKCS12=y



# Smartcard support (i.e., private key on a smartcard), e.g., with openssl

# engine.

CONFIG_SMARTCARD=y



# PC/SC interface for smartcards (USIM, GSM SIM)

# Enable this if EAP-SIM or EAP-AKA is included

#CONFIG_PCSC=y



# Development testing

#CONFIG_EAPOL_TEST=y



# Select control interface backend for external programs, e.g, wpa_cli:

# unix = UNIX domain sockets (default for Linux/*BSD)

# udp = UDP sockets using localhost (127.0.0.1)

# named_pipe = Windows Named Pipe (default for Windows)

# y = use default (backwards compatibility)

# If this option is commented out, control interface is not included in the

# build.

CONFIG_CTRL_IFACE=y



# Include support for GNU Readline and History Libraries in wpa_cli.

# When building a wpa_cli binary for distribution, please note that these

# libraries are licensed under GPL and as such, BSD license may not apply for

# the resulting binary.

CONFIG_READLINE=y



# Remove debugging code that is printing out debug message to stdout.

# This can be used to reduce the size of the wpa_supplicant considerably

# if debugging code is not needed. The size reduction can be around 35%

# (e.g., 90 kB).

#CONFIG_NO_STDOUT_DEBUG=y



# Remove WPA support, e.g., for wired-only IEEE 802.1X supplicant, to save

# 35-50 kB in code size.

#CONFIG_NO_WPA=y



# Remove WPA2 support. This allows WPA to be used, but removes WPA2 code to

# save about 1 kB in code size when building only WPA-Personal (no EAP support)

# or 6 kB if building for WPA-Enterprise.

#CONFIG_NO_WPA2=y



# Remove IEEE 802.11i/WPA-Personal ASCII passphrase support

# This option can be used to reduce code size by removing support for

# converting ASCII passphrases into PSK. If this functionality is removed, the

# PSK can only be configured as the 64-octet hexstring (e.g., from

# wpa_passphrase). This saves about 0.5 kB in code size.

#CONFIG_NO_WPA_PASSPHRASE=y



# Remove AES extra functions. This can be used to reduce code size by about

# 1.5 kB by removing extra AES modes that are not needed for commonly used

# client configurations (they are needed for some EAP types).

#CONFIG_NO_AES_EXTRAS=y



# Disable scan result processing (ap_mode=1) to save code size by about 1 kB.

# This can be used if ap_scan=1 mode is never enabled.

#CONFIG_NO_SCAN_PROCESSING=y



# Select configuration backend:

# file = text file (e.g., wpa_supplicant.conf; note: the configuration file

#	path is given on command line, not here; this option is just used to

#	select the backend that allows configuration files to be used)

# winreg = Windows registry (see win_example.reg for an example)

CONFIG_BACKEND=file



# Remove configuration write functionality (i.e., to allow the configuration

# file to be updated based on runtime configuration changes). The runtime

# configuration can still be changed, the changes are just not going to be

# persistent over restarts. This option can be used to reduce code size by

# about 3.5 kB.

#CONFIG_NO_CONFIG_WRITE=y



# Remove support for configuration blobs to reduce code size by about 1.5 kB.

#CONFIG_NO_CONFIG_BLOBS=y



# Select program entry point implementation:

# main = UNIX/POSIX like main() function (default)

# main_winsvc = Windows service (read parameters from registry)

# main_none = Very basic example (development use only)

#CONFIG_MAIN=main



# Select wrapper for operatins system and C library specific functions

# unix = UNIX/POSIX like systems (default)

# win32 = Windows systems

# none = Empty template

#CONFIG_OS=unix



# Select event loop implementation

# eloop = select() loop (default)

# eloop_win = Windows events and WaitForMultipleObject() loop

# eloop_none = Empty template

#CONFIG_ELOOP=eloop



# Select layer 2 packet implementation

# linux = Linux packet socket (default)

# pcap = libpcap/libdnet/WinPcap

# freebsd = FreeBSD libpcap

# winpcap = WinPcap with receive thread

# ndis = Windows NDISUIO (note: requires CONFIG_USE_NDISUIO=y)

# none = Empty template

#CONFIG_L2_PACKET=linux



# PeerKey handshake for Station to Station Link (IEEE 802.11e DLS)

CONFIG_PEERKEY=y



# IEEE 802.11w (management frame protection)

# This version is an experimental implementation based on IEEE 802.11w/D1.0

# draft and is subject to change since the standard has not yet been finalized.

# Driver support is also needed for IEEE 802.11w.

#CONFIG_IEEE80211W=y



# Select TLS implementation

# openssl = OpenSSL (default)

# gnutls = GnuTLS (needed for TLS/IA, see also CONFIG_GNUTLS_EXTRA)

# internal = Internal TLSv1 implementation (experimental)

# none = Empty template

#CONFIG_TLS=openssl



# Whether to enable TLS/IA support, which is required for EAP-TTLSv1.

# You need CONFIG_TLS=gnutls for this to have any effect. Please note that

# even though the core GnuTLS library is released under LGPL, this extra

# library uses GPL and as such, the terms of GPL apply to the combination

# of wpa_supplicant and GnuTLS if this option is enabled. BSD license may not

# apply for distribution of the resulting binary.

#CONFIG_GNUTLS_EXTRA=y



# If CONFIG_TLS=internal is used, additional library and include paths are

# needed for LibTomMath. Alternatively, an integrated, minimal version of

# LibTomMath can be used. See beginning of libtommath.c for details on benefits

# and drawbacks of this option.

#CONFIG_INTERNAL_LIBTOMMATH=y

#ifndef CONFIG_INTERNAL_LIBTOMMATH

#LTM_PATH=/usr/src/libtommath-0.39

#CFLAGS += -I$(LTM_PATH)

#LIBS += -L$(LTM_PATH)

#LIBS_p += -L$(LTM_PATH)

#endif

# At the cost of about 4 kB of additional binary size, the internal LibTomMath

# can be configured to include faster routines for exptmod, sqr, and div to

# speed up DH and RSA calculation considerably

CONFIG_INTERNAL_LIBTOMMATH_FAST=y



# Include NDIS event processing through WMI into wpa_supplicant/wpasvc.

# This is only for Windows builds and requires WMI-related header files and

# WbemUuid.Lib from Platform SDK even when building with MinGW.

#CONFIG_NDIS_EVENTS_INTEGRATED=y

#PLATFORMSDKLIB="/opt/Program Files/Microsoft Platform SDK/Lib"



# Add support for DBus control interface

#CONFIG_CTRL_IFACE_DBUS=y



# Add support for loading EAP methods dynamically as shared libraries.

# When this option is enabled, each EAP method can be either included

# statically (CONFIG_EAP_=y) or dynamically (CONFIG_EAP_=dyn).

# Dynamic EAP methods are build as shared objects (eap_*.so) and they need to

# be loaded in the beginning of the wpa_supplicant configuration file

# (see load_dynamic_eap parameter in the example file) before being used in

# the network blocks.

#

# Note that some shared parts of EAP methods are included in the main program

# and in order to be able to use dynamic EAP methods using these parts, the

# main program must have been build with the EAP method enabled (=y or =dyn).

# This means that EAP-TLS/PEAP/TTLS/FAST cannot be added as dynamic libraries

# unless at least one of them was included in the main build to force inclusion

# of the shared code. Similarly, at least one of EAP-SIM/AKA must be included

# in the main build to be able to load these methods dynamically.

#

# Please also note that using dynamic libraries will increase the total binary

# size. Thus, it may not be the best option for targets that have limited

# amount of memory/flash.

#CONFIG_DYNAMIC_EAP_METHODS=y



# Include client MLME (management frame processing).

# This can be used to move MLME processing of Linux mac80211 stack into user

# space. Please note that this is currently only available with

# driver_nl80211.c and only with a modified version of Linux kernel and

# wpa_supplicant.

#CONFIG_CLIENT_MLME=y



# IEEE Std 802.11r-2008 (Fast BSS Transition)

#CONFIG_IEEE80211R=y



# Add support for writing debug log to a file (/tmp/wpa_supplicant-log-#.txt)

CONFIG_DEBUG_FILE=y



# Enable privilege separation (see README 'Privilege separation' for details)

#CONFIG_PRIVSEP=y



# Enable mitigation against certain attacks against TKIP by delaying Michael

# MIC error reports by a random amount of time between 0 and 60 seconds

#CONFIG_DELAYED_MIC_ERROR_REPORT=y

More PETs for Quirky

Ttuuxxx has been compiling PETs for Quirky:

http://murga-linux.com/puppy/viewtopic.php?t=50364

You can also find my PETs here:

http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/

...these are used to build the live-CD, but not all of them are chosen. For example, Quirky 003 does not have Geany and you can grab it from here.

Mozplugger in Quirky 003

Quirky 003 has a radically different architecture for handling multimedia. There's no xine/gxine, gstreamer or mplayer.

There is ffmpeg and the mozplugger browser plugin is supposed to launch 'ffplay' to play embedded video (and maybe audio) in web pages.

I created a test html file, using the suggestion from technosaurus, posted here:
http://bkhome.org/blog/?viewDetailed=01310

Yeah, ffplay plays the video, as a separate window. I want it to be swallowed into the web page, so I edited /etc/mozpluggerrc:

video/mpeg:mpeg,mpg,mpe:MPEG animation

video/x-mpeg:mpeg,mpg,mpe:MPEG animation

video/x-mpeg2:mpv2,mp2ve:MPEG2 animation

video/mp4:mp4:MPEG4 animation

video/msvideo:avi:AVI animation

video/x-msvideo:avi:AVI animation

video/fli:fli,flc:FLI animation

video/x-fli:fli,flc:FLI animation

application/x-mplayer2:*:Windows Media video

video/x-ms-asf:asf,asx:Windows Media video

video/x-ms-wm:wm:Windows Media video

video/x-ms-wmv:wmv:Windows Media video

video/x-ms-wvx:wvx:Windows Media video

video/x-ms-asf-plugin:*:Window Media video

video/ogg:ogg:OGG stream with video

video/x-ogg:ogm,ogv:OGG stream with video

	stream swallow(ffplay) : ffplay "$file"



audio/mp3:mp3:MPEG audio

audio/x-mp3:mp3:MPEG audio

audio/mpeg2:mp2:MPEG audio

audio/x-mpeg2:mp2:MPEG audio

audio/mpeg3:mp3:MPEG audio

audio/x-mpeg3:mp3:MPEG audio

audio/mpeg:mpa,abs,mpega:MPEG audio

audio/x-mpeg:mpa,abs,mpega:MPEG audio

audio/x-ogg:ogg:OGG audio

application/x-ogg:ogg:OGG audio

application/ogg:ogg:OGG audio

audio/x-flac:flac:FLAC audio

application/x-flac:flac:FLAC audio

audio/basic:au,snd:Basic audio file

audio/x-basic:au,snd:Basic audio file

audio/wav:wav:Microsoft wave file

audio/x-wav:wav:Microsoft wave file

audio/x-pn-wav:wav:Microsoft wave file

audio/x-pn-windows-acm:wav:Microsoft wave file

audio/x-ms-wax:wax:Windows Media Audio

audio/x-ms-wma:wma:Windows Media Audio

	stream : ffplay "$file"

I then deleted /root/.mozilla/pluginreg.dat, which is recommended whenever /etc/mozpluggerrc is changed, then started SeaMonkey, opened the test html file, and YAY, embedded video!

Note that the "stream" parameter means that ffplay can take a URL directly. Without "stream", the video would be downloaded entirely first, then ffplay would be launched.

Not sure what to do about those audio files. Maybe most of those should be played with aplay?

Anyway, I want to play some more with this. Anyone know of URLs where I can test this further?...
That is, a variety of embedded formats. Or, anyone really interested and wants to setup /etc/mozpluggerrc for all situations? -- go for it!

Ntfs-3g 2009.11.14

I have updated the Ntfs-3g driver in the initrd in Woof. Previously it was version 2009.4.4.

The situation with puppies built with Woof, is you have a ntfs-3g package but that is separate from the ntfs-3g driver that I have put into the initrd. The latter is compiled statically.

This means that the live-CD has duplication, but I recently made a change in Woof so that when the live-CD is built, executables that are duplicated in the initrd and the main f.s. are deleted in the main f.s., and at bootup they get copied to the main f.s. I posted about this:
http://bkhome.org/blog/?viewDetailed=01294

I have also created a PET package with this statically-compiled ntfs-3g:
http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/ntfs-3g-2009.11.14-static.pet

Static compile howto
I compiled it with uClibc. A ready-made uClibc ext2 f.s. in a file can be downloaded from the uClibc site. You have to mount it, then chroot into it. Then compile:

# ./configure --prefix=/ --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu --target=i486-pc-linux-gnu --disable-mtab --enable-really-static
# make

secaudit.c:262:24: attr/xattr.h: No such file or directory

...I don't think I had that problem with 'xattr.h' before. I knew there is a usr/include/sys/xattr.h, so changed secaudit.c to use that, and it compiled. Hmmm.

Leeman

Ah, Leeman is your quintessential sleepy seaside village. This is where I parked, that's my car in the middle:

I was quite surprised, even though this must be still in the height of the holiday season, there was hardly anyone around.

There was a lovely cool breeze off the ocean, shady parking, change rooms with shower, and I even found a nice shaded spot on the beach. Heaven!

Leeman is big enough to have all the facilities, such as general store, fuel, hotel, telecentre, library ...oh yeah, even a tiny hardware store.

Driving around the town, I noticed quite a lot of for-sale signs. I imagine that's due to the nature of the place (seasonal holiday destination).

There was a housing boom in Perth in 2008, that went flat in 2009. Country prices followed, now they are too high I think, that's why so many for-sale signs:

http://reiwa.com.au/real+estate/leeman/

Fool that I was, I made it a day-trip only. I got back to Perenjori before the Post Office closed, and the post-mistress told me that the temperature hit 47 degrees C. It sure felt like it. I watched the weather on TV and they gave Dalwallinu as 45 degrees. Tomorrow is a "cool change", only 38.

Next time I go to Leeman, I'm going to stay overnight. Look at this nice caravan park:

http://www.leemancaravanpark.com.au/

I could have pitched my tent there for just 10 dollars:

http://www.leemancaravanpark.com.au/accommodation.html

Woof uploaded, January 5, 2010

This is the Woof version used to build Quirky 0.0.3.

More info on Woof:

http://bkhome.org/woof/

Seaside adventure

Man, it's 5am here and already so hot. I don't use air-conditioning in my house, just a fan. But when the air gets too hot, blowing hot air onto oneself just doesn't quite do it.

As the summer advances, the underground water supply pipe gradually warms, so I can't even have a cold shower, only a warm shower.

I do have one of those hole-in-the-wall reverse-cycle air-conditioners, but rarely use it. For one thing, it battles to cool even part of my weatherboard house. Then there's the 15-amp plug -- I don't have a 15-amp rated outlet.

I usually find it is better to just make use of nature. In the evening open all the windows and let the cool breeze blow through. Close the windows in the morning and use the fan when it gets a bit too hot.

The Indian Ocean is about 1.5 hours drive due West of my place, so that's where I'm heading. There's a little caravan and camping site on the coast directly West of here, named Leeman. I haven't been there for ages, can't recall what it looks like. Strangely enough, I have never before taken the direct route to the ocean from my home, always used to go to the beach at Perth.

I'm taking my tent, might stay overnight there. Or maybe not. Will just see what happens!

Quirky 0.0.3 uploaded

The latest experimental Quirky is available at ibiblio.org:

http://distro.ibiblio.org/pub/linux/distributions/quirky/quirky-003/

...it has just uploaded and not yet on any mirrors.

And there is a readme file:

http://distro.ibiblio.org/pub/linux/distributions/quirky/quirky-003/003-readme.htm

As you gather from the version number 0.0.3, this is not intended to be anything like a polished release version, or even an alpha or beta.

Have fun, play with the new ideas.

Known bug: the Ogle DVD player Stop Button in the GUI doesn't work. Gposil is working on the Ogle code for Dpup and may have resolved some issues. I just used the original code with patches from Debian.

For newcomers who are wondering what this "Quirky" thing is all about:

http://bkhome.org/quirky/

Swfdec: development slowed

I have joined the Swfdec mail list and posted about my experiences, with a request that the latest Swfdec could return to linking direct with external ffmpeg (and not with gstreamer-ffmpeg). Also sound selection used to allow 'alsa' selection, currrently doesn't (and sound doesn't work).

The main developer Benjamin responded, and basically he is focussed on developing it for his particular needs. He has Gstreamer and a particular sound setup, etc. He explained that he is not interested in maintaining anything else, and besides, development has slowed, almost stopped. Well, his words:

The reason why Swfdec does not support more than one sound system,
HTTP backend or multimedia system is very simple: The main
developer(s) (read: me) only uses one of each and wants to use his
time to improve the Flash player instead of supporting multiple
backends of a non-working player. So every backend that wasn't updated
whenever necessary was thrown out. It's not hard to add back, it's
just that the Swfdec devs don't want to keep it working.

That said, active Swfdec development has pretty much stopped, so
you'll likely not see any new features in the near future anyway.

So, to move forward with this in our puppies, one of the Puppy developers competent in C coding would have to take it on-board.

When I release Quirky 003, you can try it out. There are two main issues:

1.
I'm using Swfdec-mozilla 0.6.0 as this is required for Swfdec 0.6.8-patched. Plays videos ok, but the control buttons are weird, like if you choose to replay. Probably the latest Swfdec-mozilla has improved the user interface.

2.
I had to compile Swfdec 0.6.8-patched with an older Ffmpeg (2009-02-13) as the latest ffmpeg out of svn caused SeaMonkey to crash whenever tried to view a Flash video.

Note, if you google around, these two problems, no sound and browser crashing, are common, even with latest Swfdec and Swfdec-mozilla. There are also reports of excessive CPU usage in later versions.

I came across one person's report that the 0.6.8 version of Swfdec is the best.

Mozplugger

Gposil and I are building puppies without xine/gxine or mplayer, or gstreamer, just ffmpeg. This has been discussed in recent posts.

Without Gxine, there won't be any browser plugin, so technosaurus asked this:

One other ?? Without the gxine plugin, how is normal embedded video handled? swfdec?

like this (add to body of html template)

I think that Mozplugger might be a good solution:
http://mozplugger.mozdev.org/

It compiles easily, is very small. The tricky part will be to modify /etc/mozpluggerrc as appropriate for Quirky or Dpup.

For embedded video is seems that it can be made to swallow ffplay into the browser window. This should be interesting, I'll have a bit of a play tonight.

Pmusic: default audio player

In my latest Quirky, to become 003, Pmusic is the default audio player. So, I have placed the script 'pinstall.sh' into the Pmusic PET package:

#!/bin/sh



if [ "`pwd`" != "/" ];then



 echo "Configuring Pmusic..."

 

 if [ ! -e ./usr/bin/aqualung ];then

  echo '#!/bin/sh' > ./usr/local/bin/defaultaudioplayer

  echo 'exec pmusic "$@"' >> ./usr/local/bin/defaultaudioplayer

  chmod 755 ./usr/local/bin/defaultaudioplayer

 fi



fi

I have created 'pmusic-0.9.9-1.pet' and uploaded to the ibiblio.org PET repository.

Zigbert, if you read this, is it ok for you to put this script in future Pmusic PETs? Otherwise, I'll manually insert it each time I upgrade Pmusic.

Woof uploaded, January 3, 2010

I have just uploaded the latest Woof.

One thing that I have done is remove Shinobar's 'pupsaveconfig'. After using it for awhile, I feel that there are too many questions and steps at shutdown. Also, as I run my puppies from live-CD and do a lot of rebooting, I got very annoyed having to reinsert the CD and press ENTER key each time.

I can conceive of a simpler way of implementing an X GUI for the pupsave, so will revisit this later.

Interesting, this was my first need to use 'bones diff' to compare the commits where Shinobar's changes were introduced, and I was easily able to restore the previous files.

Ffmpeg recompiled for 003

Here is a record of what I have compiled, that will (probably) be in Quirky 003:

i had a problem, after compiling ffmpeg 20091229 and swfdec-mozilla, seamonkey crashes when try play any flash videos (using Swfdec 0.6.8-patched and Swfdec-mozilla) 0.6.0 -- see recent posts).

so, staying with the ffmpeg from t2 (as used in Quirky 001 and 002)...
i want to recompile it with 'ffplay' which requires SDL...

SDL 1.2.14
# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu --disable-esd --disable-pulseaudio --disable-arts --disable-mintaudio --disable-video-cocoa --disable-video-directfb --disable-video-opengl --enable-nasm --disable-video-photon --disable-video-carbon --disable-video-ps2gs --disable-video-ps3 --disable-video-ggi --disable-video-vgl
# make
# new2dir make install

amrnb 7.0.0.2
# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu --with-downloader=wget
# make
# new2dir make install

amrwb 7.0.0.3
# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu --with-downloader=wget
# make
# new2dir make install

speex 1.2rc1
# ./configure --prefix=/usr --localstatedir=/var --sysconfdir=/etc --build=i486-t2-linux-gnu
# make
# new2dir make install

yasm 0.8.0
x264 needs this to compile. ffmpeg can also use it.
# ./configure --prefix=/usr --localstatedir=/var --sysconfdir=/etc --build=i486-t2-linux-gnu --disable-python --disable-debug
# make
# new2dir make install

x264 20091230
tricky, installs to /usr/local, have edited 'configure', changed prefix='/usr'
note, configure needs 'join' command, from coreutils pkg, not in quirky.
hmmm, it ignores the --host=i486 instruction, compiles for i686...
# ./configure --enable-shared --host=i486-t2-linux-gnu --enable-pthread
# make
# new2dir make install
...note, the executables got linked statically with the libx264 library, so have removed them from the PET package, just kept the shared lib as that is what ffmpeg needs.
...note, new2dir failed to grab the symlink /usr/lib/libx264.so
...crap, tried to use in ffmpeg, but configure ran a test and reported 'undefined reference to `x264_encoder_open''

x264 20090113
compiled as above. this time it is not compiling for i686!
now for the 64,000 dollar question, is ffmpeg happy? ... yes

ffmpeg 2009-02-01
removed --enable-libvorbis, cos libvorbisenc is 1MB. nah, back in...
# ./configure --prefix=/usr --cpu=i486 --arch=i486 --enable-libmp3lame --enable-libfaac --enable-libfaad --enable-pthreads --enable-small --enable-postproc --enable-swscale --enable-gpl --enable-shared --enable-nonfree --enable-x11grab --enable-bzlib --enable-libamr-nb --enable-libamr-wb --enable-libspeex --enable-zlib --disable-debug --enable-libvorbis --enable-libx264
# make
# new2dir make install

...works! swfdec flash player works in seamonkey. ffplay works.

Notes
The latest ffmpeg from svn almost plays RealMedia videos -- I got video, and although it does have the 'sipr' audio codec it doesn't work. On the other-hand, the ffmpeg version above cannot play my RM files at all.

I really need to look into getting the latest Swfdec working (0.6.8 is an old version) and I intend to communicate with the author about this.

Then I'll be able to upgrade to the latest ffmpeg.

My recent posts on Swfdec:
http://bkhome.org/blog/?viewDetailed=01298
http://bkhome.org/blog/?viewDetailed=01299

Multimedia unhappiness

It took me all morning, I compiled all of these:

SDL 1.2.14
opencore-amr 0.1.2
speex 1.2rc1
xvidcore 1.2.2
yasm 0.8.0
x264 20091230
ffmpeg 20091229
liboil 0.3.16
libsoup 2.28.2
swfdec 0.6.8-patched
swfdec-mozilla 0.6.0

...as well as some other stuff, such as Ogle.

ffplay works nice.
But, whenever I try to view a Flash video, SeaMonkey crashes! Oh dear.

Ogle DVD player

We have been discussing a major rethink about multimedia packages in our puppies:

http://bkhome.org/blog/?viewDetailed=01304

I downloaded Ogle and Ogle_gui (GTK2 application) from the project home, but I found both to be buggy. But yes, they do work, and very light-weight.

But I have just now found that the Debian guys have been patching both ogle and ogle_gui, right up until 2008:

http://mirror.aarnet.edu.au/debian/pool/main/o/ogle/
http://mirror.aarnet.edu.au/debian/pool/main/o/ogle-gui/

So, it's looking like Dpup and Quirky are undergoing major changes in the multimedia arena!

Tonight I'm going to go through and recompile ffmpeg, without x264 only xvid -- as tempestuous suggested. I'll dump xine and gxine. Ogle_gui for playing DVDs, ffplay for video files. The Adobe Flash player is also gone, replaced with Swfdec. This will all be in Quirky 003, which I hope to upload in a couple of days.

Dpup
If you want to see ffplay in action as a media player, gposil has just released Dpup 484beta1:

http://www.dpup.org/downloads.html

Fundamental questions about multimedia packages

There is discussion about .3g2 video files playing video but no sound:

http://murga-linux.com/puppy/viewtopic.php?t=50409

In Quirky I have just been through the exercise of recompiling ffmpeg (from svn, December 29), with libopencore-amr, x264 and xvid support, as well as all the usual packages in Puppy/Quirky.

But, I was very disappointed when I played the .3g2 files in Gxine and still no sound.

Then I tried 'ffplay', a console video player, and hey, I get sound!

Gposil has been through the same exercise for Dpup, and has also commented how good ffplay is. Incidentally, note that ffplay needs libSDL, so there is a bonus for games if we decide to put ffplay into our puppies.

Anyway, all of this is making me rethink the fundamentals. I've got Swfdec-mozilla compiled against ffmpeg and doing a pretty good job as a Flash player plugin. Well, here are the questions:

1. x264, xvid
I am not really clear on why we need these dependencies in ffmpeg. In my general rambling around on the Internet, ffmpeg seems to handle everything, well most files, without these dependencies. What files/streams are going to need these dependencies? In other words, is it really important to have ffmpeg compiled with these dependencies?

2. Why do we need xine?
If ffplay can do such a good job at playing videos, and something else such as Pmusic can play audio files, and pCD play audio CDs (I don't know about DVD videos though), why do we need xine and gxine?

Regarding question 2, couldn't we develop our own wrapper for ffplay? Ffplay can even toggle full-screen:
http://linux.die.net/man/1/ffplay

Cabernet Shiraz

"Mr C" (who wishes to remain anonymous) sent me a bottle of Scotts Creek Cabernet Shiraz for Christmas -- but it arrived today, December 31. AU50 dollars cash as well.

That's great, I'll have a glass (or two) this evening. There are a couple of nice musical programs on ABC tonight, so I'll settle down in front of the TV, glass in hand.

Woof uploaded, December 31, 2009

This has various improvements, including shinobar's X GUI 'pupsave' script.

Further information on Woof:
http://bkhome.org/woof/

'pupsave' X GUI

Shinobar has posted to the forum Quirky 002 feedback thread:

quirky-002 does shudown by power button if acpid is fixed.
but still has a problem if it is boot up without pupsave(quirkysave).

boot up without pupsave (asume acpid is fixed), and pressed power button, what happens?
after 30 seconds going to shutdown, then the dialog asking where to save the session appears and waits reply forever.

one solution is add '--timeout' option with the dialog.
another way is use pupsaveconfig, a GUI tool to setup where to save the session.

i prepared the pupsaveconfig of the puppy 4.3.1JP ready for quirqy-002.
find pupsaveconfig-1.8.pet available from here: http://shino.pos.to/linux/puppy/

pupsaveconfig-1.8.pet has /etc/rc.d/rc.shutdown in it, with enhancements to recognise if the pupsave has already been chosen by the X GUI. I have mofified rc.shutdown since 002, with a fix to recognise memory cards that have /dev/mmcblk* device nodes. I have merged my lasted fix with shinobar's script and placed it at /etc/rc.d in Woof.

The scripts in pupsaveconfig-1.8.pet are pretty integral to the 'rootfs-skeleton' in Woof, so I thought it best to put them directly into the skeleton f.s. in Woof rather than having them as a separate package. The 'ja' locale is also in Woof.

Ok, that is done, shinobar's 'pupsaveconfig' is now builtin to Woof. I'll upload the latest Woof soon.

acpid fix

On the forum Quirky 002 feedback thread, shinobar reported:

Quirky-002 does not shutdown by power button because acpi_poweroff attempts to call /usr/X11/bin/wmpoweroff but is not there.
Rewright '/usr/X11/bin/wmpoweroff' to 'wmpoweroff' at the bottom of the script /etc/acpi/actions/acpi_power_off.

Or, install acpid-1.0.8-7.pet available from here: http://shino.pos.to/linux/puppy/.
The duration of the closing dialog is adjustable by changing the number 'TIMELIMIT=30' at the top of /etc/acpi/actions/acpi_power_off.

I will upload this PET to the Quirky repository at ibiblio.org soon.

Swfdec-mozilla Flash plugin

I have uploaded liboil 0.3.16, libsoup 2.28.2, libgpg-error 1.7, libgcrypt 1.4.5, gnutls 2.8.5, swfdec 0.6.8 (patched) and swfdec-mozilla 0.6.0 sources:

http://bkhome.org/sources/alphabetical/

To compile the Swfdec player with least dependencies, build as follows:

liboil 0.3.16
# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu
# make
# new2dir make install

libsoup-2.28.2
disable ssl, compile static lib only...
# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu --enable-debug=no --disable-shared --enable-static --without-gnome --disable-ssl
# make
# new2dir make install

swfdec 0.6.8-patched
# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu --with-audio=alsa --disable-gstreamer --enable-ffmpeg --disable-static --enable-mad
# make
# new2dir make install

swfdec-mozilla 0.6.0
# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --build=i486-t2-linux-gnu --with-plugin-dir=/usr/lib/mozilla/plugins --disable-static
# make
# new2dir make install

Libsoup is compiled statically into Swfdec as it is somewhat crippled without the SSL support. I could have compiled liboil statically, in fact I could even have compiled Swfdec statically into Swfdec-mozilla for the absolute smallest size.

Gstreamer option
There is also the option of abandoning Xine and going over totally to Gstreamer. The main problem with that is Gstreamer does not want to compile with external ffmpeg (well, perhaps that could be hacked, but the Gstreamer developers very strongly recommend to use their internal ffmpeg). Puppy has some apps that rely on the external ffmpeg.
It would be interesting to get Gstreamer to compile with external ffmpeg and see how well it works. Swfdec can be compiled against Gstreamer instead of ffmpeg.

Oh yes, I should report on that. I did compile Swfdec against Gstreamer, and some videos at youtube.com just were a black window -- no video or sound.
On the otherhand, for Swfdec 0.6.8 with ffmpeg, so far I've tested about a dozen videos at youtube.com and they all work.

Swfdec Flash player

I compiled Swfdec version 0.8.4 and Swfdec-mozilla 0.8.2. The latter is a Flash plugin for Mozilla-based browsers.

It plays most of the videos at youtube, except for one small problem -- no sound. Running SeaMonkey from a terminal, I get this error message:

SWFDEC: ERROR: swfdec_audio_decoder.c(232): swfdec_audio_decoder_errorv: error decoding audio: no suitable decoder for audio codec 10

I have done a quick google and this error is reported many times. Have not yet seen a solution, will investigate some more.

Memtest in Linux

panzerpuppy wrote:

How about including Memtest (accessible from the boot menu) in the next build of Quirky?
I really miss this useful app. Great for memory diagnostics and checking the bandwidth/performance of the memory and CPU caches.

The 2.6.26 and later kernel has memtest builtin, that can be invoked at the boot prompt. See my earlier report:

http://bkhome.org/blog/?viewDetailed=01142

The 2.6.30.5 kernel used in Puppy 4.3.1 is configured with memtest enabled (as is the 2.6.31.5 kernel in Quirky), but I was unable to get anything in 'dmesg' despite what Dougal investigated should be happening.

Obviously I was doing something wrong when I tried to activate it. Maybe there was something in 'dmesg' and I overlooked it?

I wish that I could get this issue resolved. Every now and again someone posts that they would like memtest to be in Puppy, I respond that it is already in the kernel, and there the matter stops, noone resolves it. Would someone take this onboard and solve this mystery? There is probably a very simple solution.

User Agent Switcher for SeaMonkey

sage and infectiia have complained that SeaMonkey does not work on the NatWest Bank website.

I have posted a query about this here:
http://murga-linux.com/puppy/viewtopic.php?p=375342#375342

If the 'User Agent Switcher' SM addon fixes it, I'll make it into a PET package and build the next Quirky with it.

UPDATE:
I looked at www.nwolb.com (NatWest online banking) and it told me I had the wrong browser. I then installed User Agent Swicher, restarted SM, chose "Internet Explorer 7" (see Tools menu), then tried that site -- this time success. Couldn't login of course.

Comment: What a crappy bank, blocking SM. If I had an account there, I would vote with my money, go elsewhere and let them know why.

Pnethood bugfix

I was looking through the Quirky 002 feedback thread and thanks to shinobar who posted a fix for Pnethood:

http://murga-linux.com/puppy/viewtopic.php?t=50319&start=30

I have updated the package, so the next build of Quirky will have the fix.

Woof uploaded, December 26, 2009

I found myself with a couple of idle hours yesterday, so opened up my laptop and did a little bit of work on Woof.

There was something I was meaning to do for a very long time, and finally have done it.

In the Puppy ISO there is some duplication of binary excutables in the 'initrd.gz' and 'pup-nnn.sfs' main filesystem file. This adds to the size of the ISO.

What I have done is the '3builddistro' script in Woof deletes the duplicate files in the main f.s., except for 'busybox' and 'cp' -- the busybox in the main f.s. is different and cp is needed later by the build script. These are the executables that are deleted in the main f.s.:

guess_fstype modprobe lsmod fusermount e2fsck disktype e3 elspci resize2fs fsck modinfo rev find waitmax hotplug2stdout_notimeout losetup ntfs-3g

At bootup, the 'init' script copies all of the above from the initramfs to the main f.s., at their correct locations. 3builddistro creates bin/TARGETEXES in the initramfs that lists the executables as well as their correct path in the main f.s., and the init script reads this file.

Anyway, the end result is surprising, the ISO file is 0.5MB smaller -- all those little apps do add up. This is good, it applies to all puppies/puplets/quirklets built with Woof.

I have just uploaded this latest Woof.

Don't think I'll be doing anything on Woof/Puppy/Quirky today, got other things to do!

Woof uploaded, December 24

It is early morning on December 24 here and I am just having a last look on the Internet before heading off for my Xmas break.

Good thing too, I found this report by playdayz:
http://murga-linux.com/puppy/viewtopic.php?t=49411

I don't know about the Python package problem with Upup (or as some people prefer, uPup!), but I have fixed the entries in file 'Packages-puppy-woof-official'.

I have uploaded Woof just now, with the fix. Playdayz, thanks for that.

Ah, I saw on the news, some of you guys in the US and Europe are snowed under. Today we are having a mild 30 degrees C, tomorrow 36. I'm taking my bathers.

Woof uploaded December 23, 2009 (again)

As explained in recent blog posts, there is a problem in some cases with hotplugging of memory cards, with the deskop drive icons not appearing (or disappearing).

See recent posts:
http://bkhome.org/blog/?viewDetailed=01291
http://bkhome.org/blog/?viewDetailed=01289

I had a solution, that used 'dd' to read from removable USB drives every two seconds, however I wasn't happy with it. I don't like the idea of a plugged-in card being read every two seconds.

So, I have worked on /sbin/pup_event_frontend_d and now have it so 'dd' only runs when a card is not plugged in. It works nicely, except that removal of a card is now not properly detected, meaning that an icon may remain on the desktop after a card is unplugged -- however I modified /usr/local/bin/drive_all to detect an invalid icon if it is clicked on and it is then removed.

This is a workable compromise, with low overheads, and most inportantly I think doesn't keep reading from a card every 2 seconds to see if it is still there.

Actually, the problem is at the kernel and kernel driver level, and it might go away as the kernel keeps improving. For example, external drive sde, and I plug in a card, /sys/block/sde/sde1 directory appears -- but when I unplug the card that directory remains indefinitely -- until an attempt is made to read from that drive (as with 'dd').

Anyway, I have uploaded Woof with this latest solution.

PCMCIA/MMC/SD support at bootup, hotplug drives

Improved PCMCIA/MMC/SD recognition in initramfs
This was discussed here:
http://bkhome.org/blog/?viewDetailed=01289

I think that I have got it so that the 'init' script will recognise a much wider range of memory cards, so they can be used to hold a 'pupsave' file or even booted from (if the BIOS supports it -- unfortunately my Acer Aspire laptop can't boot from the inbuilt multi-function card slot -- which would have been so nice -- I reckon that's something to look for if you are shopping for a new laptop).

Desktop drive icons hotplug support
The above link has discussion with pakt about a limitation, where a memory card may be plugged or unplugged but there is no desktop drive icon, yet Pmount does recognise it.

I've got an external USB multi-function card reader that has this problem. When plugged in, /sys/block shows that it has sdc, sdd, sde and sdf (in my case), regardless whether any card is plugged in or not.

Even if I plugin a card, /proc/partitions does not update.

The only way to detect that a card is inserted is to actually probe it. The 'probedisk2' and 'probepart' scripts, that are called by Pmount, do this -- they use 'dd' to try and read from the card -- this causes /proc/partitions to update.

I have put this probing into /sbin/pup_event_frontend_d, but in an extremely restrained way, as the scan occurs every 2 seconds and we don't want an overhead that will noticeably impact on performance -- that is, we don't want the user to notice any degradation.

It is working for me. I can plug and unplug an SD card in my multi-function adaptor and the desktop drive icon appears and disappears accordingly.

Woof uploaded, December 23 2009
I have just uploaded Woof with the above fixes.

Those unfamiliar with Woof can read more here:
http://bkhome.org/woof/

I'm going to be taking it easy for the next 5 - 6 days. Have a nice Christmas holiday everyone! For those who don't have a holiday at this time of year, best wishes also and take it easy too!

More PETs for Quirky

Ttuuxxx has already started compiling PET packages for Quirky:

http://murga-linux.com/puppy/viewtopic.php?t=50364

Of course, older Puppy 2, 3, 4 PETs will also likely work.

Note, the packages used to build Quirky were recently compiled in T2 so all the libraries are very recent versions.

SD card support at bootup

A problem with Puppy is that an internal MMC or SD card is not recognised at bootup. It is recognised after bootup. This means that you can't use it to boot from or to hold a 'pupsave' file.

By "internal" I mean laptops that have a card reader built-in. A card reader that connects via USB or IDE interfaces is different.

I have fixed this on my laptop, which has a Texas Instruments multi-function card reader interface. I presume this is what most laptops would have?

If you do have a laptop with internal card reader, could you confirm something for me. Run this:

> elspci -l

...and see if you get a line with "018000" in it. If so, we're in business.

If not, run:

> lspci
or
> scanpci

and see if you can identify which line from elspci corresponds to your card reader interface.

Oh yeah, the kernel modules. On my laptop, these are loaded:

tifm_sd
mmc_block
mmc_core
tifm_7xx1
tifm_core

Run 'lsmod' to find out what is loaded in your laptop. I'm interested in knowing if your hardware needs some other modules.

Quirky 0.0.2 uploaded

I made a very brief announcement here:
http://distro.ibiblio.org/pub/linux/distributions/quirky/quirky-002/002-readme.htm

The desktop features "Tillman the skateboarding bulldog".

Download:
http://distro.ibiblio.org/pub/linux/distributions/quirky/quirky-002/
http://ftp.nluug.nl/ftp/pub/os/Linux/distr/quirky/quirky-002/

Build 002 is for sanity testing my work on reducing size, so testers welcome. I would like to know if any apps are not working. Look around the menu, there are new things there.

002 is my Christmas present. Have a nice time everybody. Between now and Christmas I'll be working on a few things -- like, a few people want booting from SD card (with a laptop inbuilt SD card slot), so I might take a look at that. Then a break for a few days, family get togethers, etc.

Woof, December 21, 2009

I have just uploaded the latest Woof.

To download it, you need to install Bones version 0.1 (which is already in Quirky 002). Read about Bones, also has download link:
http://bkhome.org/bones/
...also explains how to download Woof.

If you need to catch up on what Woof is all about:
http://bkhome.org/woof/

I will be announcing Quirky 002 soon, built from Woof. This build is pretty much a straight Tpup, but at 88MB a pretty compact starting point for anyone wanting to develop their own Tpup.
Like for example, you could build it with the 2.6.30.5 kernel and all the extra drivers that are bundled with it (like lots of analog modem drivers).

Modules loading preferences fix

Rerwin (Richard) has posted this message to the Puppy 431 bugs thread:

Reports of unreliable handling of module preferences, as given in the BootManager Preference option, probably resulted from a bug in the pup_event_backend_modprobe script, which is responsible for the loading of appropriate hardware driver modules. The attached package, pup_event_backend_modprobe_fix_to_p43x-1.pet, resolves that bug and extends the capabilities of the Preference function to (1) provide predictable control over module preferences and (2) support some special cases of module selection and substitution.

The current implementation of preferences not only can fail to make a substitution, but may load a module other than the one specified, if more than one alternative is appropriate. Specifying the name of a module as a substitute merely results in substitution with whatever module is available for the device; if there are multiple candidates available, the one loaded may not be the one named as the preference.

This new version of the script extends the preference function by (1) allowing multiple preferences for the same module (separated by a "|" and ordered first-choice-first if needed), the selection being governed by the particular modules that identify themselves as supporting a particular device, and (2) allowing dynamic activation of a set of preferences needed when certain conditions occur. The special preferences reside in /etc/rc.d/MODULESCONFIG as assignment statements setting variables named PREF...; they are not accessible through the BootManager.

One use of these extensions is for hybrid USB storage-modem devices (e.g., wireless modems), which require two drivers but only one of them normally gets loaded. All such devices may initially require module usb-storage, then need one of several possible modem drivers. If the normally-loaded driver is usb-storage instead of the others, an internally activated preference is used if usb-storage is already loaded (which it normally is), but the modem driver must be the one(s) intended for the modem (known at load time).

Another use is for supporting ALSA modems that might instead actually be Conexant modems. The Conexant (Linuxant) driver issues an error message if the modem is not a Conexant. In this case, preferences can be activated to force use of the alternate (ALSA) driver after a reboot of Puppy. The new features could support other special module-loading situations as they become known.

This package/dotpet has been tested with all three kernel versions of Puppy 4.3.1. It should be installed by anyone experiencing a problem with preferred modules not loading (e.g., atl1e instead of atl1c). Do not use this package on Puppies 4.1.x or 4.2.x; instead, use the similar, but bugfix-only package created for those Puppies and available in the Bugs/Fixes threads for 4.1.2 and 4.2.1. Use of the HSF/ALSA reselection requires installation of the new modem "fix pack" package (to be uploaded soon in this thread). This fix can be applied to any Puppy 4.3 installation and is a candidate for inclusion in 4.4CE.

I have applied rerwin's fixes to Woof. The affected files are /sbin/pup_event_backend_modprobe, /etc/rc.d/MODULESCONFIG and /etc/rc.d/rc.sysinit.

I will probably upload the latest Woof later today.

Gxine rollback

I had fun and games with Gxine last night. I found that Gxine 0.5.904 (compiled in T2) crashes with my new cutdown SeaMonkey 1.1.18. Ah, SM 2.0 was compiled in T2 and Gxine uses its libmozjs, so I recompiled Gxine 0.5.904 against the SM 1.1.18 libmozjs ....oh, it still crashes.

So, I compiled Gxine 0.5.9 against my SM 1.1.18, and that works.

Interesting, Gxine 0.5.904 was stable when compiled against SM 2.0.

Quirky 002 soon

I have updated Pburn to 3.1.7, Precord to 2.0.6 and NicoEdit to 2.5.1.

Or rather, I made a PET package for the latest NicoEdit, but I have decided to build Quirky 002 with only Geany.

I have made a little short-list for Quirky 002, which I think I can get through tonight and tomorrow morning, so maybe will upload 002 tomorrow afternoon.

It's going to be a "straight Tpup", normal desktop with JWM and tray on the bottom, usual icons on the desktop. I do intend to play some more with a cleaner desktop, however the purpose of 002 is to test that my efforts to reduce the size still results in a sane system.

So, try all the apps, printing, scanning, whatever, let me know if anything is broken. Note, I have not yet put in Jemimah's Pwireless2 -- probably will do that for 003.

Bones 0.1

Bones is getting a bit more meat on it, and I have now released version 0.1.

This is a PET package, available from here:

http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/bones-0.1.pet

Thanks to ecube (Olov) who has contributed a helper-script, bones_diff.pl, Bones is now at the early stage of being able to examine differences between saves.

There is now a main GUI window, invoked by typing 'bones' at the prompt, so Bones is heading towards being highly GUI-fied. I have updated the web page with information:

http://bkhome.org/bones/

Does anyone know an easy way to extract just one file out of a .tar.gz, without actually expanding the whole thing? I think Midnight Commander can do it, but I need something standalone for that one purpose only.

pupRadio and pupTelly

It seems like just about every day I visit the forum and there is some new exciting project started!

pupRadio and pupTelly is an example, created by 01micko:
http://murga-linux.com/puppy/viewtopic.php?t=50016

Puppy 431 has Pstreamvid created by trio:
http://murga-linux.com/puppy/viewtopic.php?t=45023&search_id=1526069224

One thing I was wondering, trio modified his script to work specifically with Gxine 0.5.9, but my Quirky has Gxine 0.5.904.

What about pupRadio/pupTelly, is it happy with either version of Gxine? Of course, your script can read the version of Gxine by

gxine --help | grep '^gxine ' | cut -f 2 -d ' '

...and then behave accordingly.

Proposed NicoEdit 2.5.1

I posted a request to this forum thread:
http://murga-linux.com/puppy/viewtopic.php?p=373513#373513

My forum post:

Nicolas, the developer of NicoEdit, is busy with other things right now, so not currently keen to work on NicoEdit.

Version 2.5 fixed a problem, the PPLOG script did not open. It does not fix the shebang line '#!/bin/ash' detection that I had requested. Also, 2.5 does not have the menu option to manually choose color highlighting -- which 2.4 does have.

So, is anyone interested in examining 2.4 and 2.5 source and put the menu entry for syntax highlighting back in? -- because NicoEdit does not always autodetect the type of file.

Also, figure out how to add '#!/bin/ash' to the other shebang lines '#!/bin/sh', '#!/bin/bash'.

Then we will have say version 2.5.1.

Anyone interested in having a play with Genie coding?

NicoEdit source:
http://bkhome.org/sources/alphabetical/n/
User: pup�py Pass: lin�ux

Precord 2.0.4

I introduced Precord recently:
http://bkhome.org/blog/?viewDetailed=01261

Forum:
http://murga-linux.com/puppy/viewtopic.php?t=49907

Version 2.0.4 is very nice, and will be in Quirky 002. I have a problem with sound being too quiet on my laptop, and I appreciated being able to adjust levels on the mixer and see immediately what sound-level I got -- I found "mic boost" settings that got my recording level high enough.

A really nice little app!

pCD 1.1, gtkdialog3 rolled forward

I have placed zigbert's pCD version 1.1 audio CD player into Quirky.

I had previously announced gtkdialog3 patched by Patriot:
http://bkhome.org/blog/?viewDetailed=01264
But ipinfo was broken so I rolled back:
http://bkhome.org/blog/?viewDetailed=01267

However, I have communicated with Patriot, and the ipinfo script is the one that is wrong. Extract from pm from Patriot:

I've taken a look at ipinfo1 and yes, ipinfo1 uses the very method (the bug) that zigbert wishes to be fixed to exit the program ...

The documented method to exit is exit:exit ... The closewindow:x method was supposed to close an opened child window (and now working correctly in the patched gtkdialog3) ... My suggested remedy for the ipinfo1 :

Code:

Or, simply omit the line, as by default gtkdialog will exit if no action is defined.

So, I have fixed ipinfo and rolled gtkdialog3 forward again.

HomeBank 4.1

There is a new version with lots of bugfixes:

http://homebank.free.fr/

I have compiled it for Quirky.

I did not compile-in libofx support, as the 'libofx' package needs 'opensp' which would not compile.

Floppy Formatter 1.7

Forum member 8-bit has incorporated support for both internal and USB floppy drives into his Floppy Formatter package. This is now version 1.7:

http://www.murga-linux.com/puppy/viewtopic.php?p=367386#367386

I have added this to Quirky.

Internet Connection Wizard and ipinfo

I have incorporated the 'ipinfo' script into the Internet Connection Wizard, callable by a button.

gHasher

A small application to calculate md5 (and other) hashes of files. Discovered by ttuuxxx. I have compiled this for Quirky, and the PET is only 13KB.

gHasher home:
http://asgaard.homelinux.org/code/ghasher/

Free SoftMaker Office 2008

For Windows only though. If you are interested, get it from here, but note that the offer expires on December 31:

http://loadandhelp.com/home-en.html

Ayttm 0.6.1

I have compiled Ayttm version 0.6.1 for Quirky. This has support for Yahoo webcam (using the jasper package).

Project changelog:
http://ayttm.git.sourceforge.net/git/gitweb.cgi?p=ayttm/ayttm;a=blob;f=ChangeLog;h=9e090ace2c35a722d2dc4a301215e79cef359e38;hb=HEAD

Hmmm, The PETs for Ayttm and Jasper are about 0.9MB, so if I went for Meebo I could save that much off the ISO. But, I am not inclined to move Quirky toward online applications, not yet anyway.

Regarding release of Quirky 002, I'll aim to get it out before Christmas, so consider it as my Christmas pressie to ya all.

Quirky 002 coming soon

In a few days probably, and I'll upload Woof as well.

I have just built the precursor to 002, and the ISO file is 87.6MB. Running it now.

This has my new tiny SeaMonkey. Noticeably faster startup.

Compiling SM 1.1.18

The SM 2.0 PET package is 11.33MB, that of SM 1.1.18 is 10.28MB. As one of the aims of Quirky is to become smaller, I am planning to roll SM back.

Right now I'm recompiling SeaMonkey 1.1.18, trying some different configure options to see if I can get it even smaller.

NicoEdit 2.5

Nicolas has made a couple of small improvements and released version 2.5.

One of those is to recognise the "shebang line" (first line in a script) of "#!/bin/ash"

Another was 2.4 refuses to open the PPLOG Perl script. I haven't tested yet, but that should be fixed now.

2.5 source:
http://bkhome.org/sources/alphabetical/n/nicoedit-2.5.tar.gz

PET package compiled in Quirky:
http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/nicoedit-2.5-q1.pet

Searchmonkey is back

Searchmonkey is the best file and file-content finder, but in Puppy4 it crashes, which is why it wasn't chosen.

In Quirky however, it works, but then I discovered that the project has been revitalised, and there is a new version, 0.8.1.

Searchmonkey home:
http://searchmonkey.embeddediq.com/index.php

My PET, compiled in Quirky:
http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/searchmonkey-0.8.1-q1.pet
...PET size is 92KB.

I have chosen Searchmonkey as the sole file finder app in Quirky.

USB Floppy Formatter

Forum member 8-bit has developed this:
http://murga-linux.com/puppy/viewtopic.php?t=48053

I have repackaged it slightly and uploaded to here:
http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/usb_floppy_format-1.5-p4.pet

Note that the package has 'ufiformat' in it which is a binary executable (compiled in Puppy4).

Ufiformat home page:
http://www.geocities.jp/tedi_world/format_usbfdd_e.html

IPInfo network information

Vovchik has created a nice little GUI application that displays information about the network:
http://www.murga-linux.com/puppy/viewtopic.php?t=47262

Some of the information was not displaying properly in the tabs, so I hacked the script a bit to fix it. I also added a "Please wait" message at startup.

I have placed IPInfo into Woof, at /usr/sbin/ipinfo and created a .desktop file so there is a menu entry.

Then I noticed that the 'Cancel' button did not work. This is due to Patriot's patched gtkdialog, so I have rolled back to 'gtkdialog-0.7.20-1.pet'.

NTFS-backup

Kirk created a tiny little app that is handy for Windows people. Extract of kirk's message to me:

It's a GUI for ntfsclone, it's written with xdialog. I've used it many times with Windows 2000, XP, and Vista. Ntfsclone is included in Puppy and it's very fast. My kids hose up their laptops with Vista a lot, with ntfsclone it takes about 15min to restore the c: partition.

I have uploaded the PET package here:

http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/ntfs-backup-1.pet

...the PET is 2.1KB!

Pfind 4.16, Pburn 3.1.6

I have updated zigbert's apps in Quirky. The PETS will be uploaded soon to:

http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/

gtkdialog 0.7.20-patriot-e-1

Zigbert recently posted the latest gtkdialog sources that had been patched by Patriot. I think that Patriot had also included patches from Ubuntu.

I have uploaded this source to here:
http://bkhome.org/sources/alphabetical/g/gtkdialog3-0.7.20-patriot-e-1.tar.gz

I have compiled it in Quirky/Tpup and uploaded the PET to here:
http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/gtkdialog3-0.7.20-patriot-e-1.pet

The forum thread where this came from:
http://www.murga-linux.com/puppy/viewtopic.php?p=357567

dhcpcd 5.1.3

I have upgraded the package in Quirky. Configuration:

# ./configure --libexecdir=/lib/dhcpcd --dbdir=/var/lib/dhcpcd --sysconfdir=/etc

...note, set to /var/lib instead of /var/db.

gtrayicon_network

Jemimah uses a patched Gtrayicon v1.1 in Pwireless2. I decided to package the former as a separate package and named it 'gtrayicon_network' -- a dumb name, but I've done it now.

The Gtrayicon home page:
http://gtrayicon.sourceforge.net/

I applied patches by Jemimah and technosaurus from here:
http://www.murga-linux.com/puppy/viewtopic.php?t=48816

I uploaded the patched source here:
http://bkhome.org/sources/alphabetical/g/gtrayicon_network-1.1.tar.gz

I uploaded the PET package here:
http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/gtrayicon_network-1.1-q1.pet
...size of PET is 22KB.

What I aiming for is to pull all of the binary executables out o Jemimah's Pwireless2 package.

Precord, Woo-ff, mhWavedit, Pmusic, Asunder

As I put acpid into Quirky, I wanted to take something out, as one objective for Quirky is to be always moving down in size.

Hmmm, the acpid PET package is only 12KB. Anyway, I did have in mind something to take out, so have done this...

I removed Sweep (254KB PET) and replaced it with mhWavedit (172KB PET).

Then I put in Precord 1.0.2, a sound recorder created by mcewanw:
http://murga-linux.com/puppy/viewtopic.php?t=49907
...the PET is only 3.4KB.

I took out Pawdioconverter (has missing Sox dependency, and 12KB PET) and I added Woo-ff 0.1.1, an audio file converter, created by technosaurus:
http://www.murga-linux.com/puppy/viewtopic.php?t=45680
...the PET is only 2.6KB!

The I added zigbert's Pmusic 0.9.9:
http://www.murga-linux.com/puppy/viewtopic.php?t=31206
...the PET is 33KB.

I replaced Ripoff CD ripper (50KB) with Asunder (45KB).

...overall I'm down by 15KB!

Remaster-CD script updated

Shinobar suggested this for 4.4:

http://www.murga-linux.com/puppy/viewtopic.php?t=46941&start=180

So, I have put this 'remasterpup2' script into Woof.

I know that I have been dragging my feet with regard to adding locale support to my scripts, but shinobar has done so much work on this script, so it is now in Woof -- shinobar has also made improvements, like choice to get the Puppy files from CD or elsewhere.

acpid

This looks pretty desirable for laptops. Forum member Patriot has done all the work, see forum:

http://www.murga-linux.com/puppy/viewtopic.php?t=38173

Shinobar has been tweaking it for the Japanese Puppy and has PETs available here:

http://shino.pos.to/linux/puppy/

I have decided to put it into Quirky, so I have uploaded the latest from the Japanese Puppy to my Quirky PET repo:

http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/acpid-1.0.8-6-p4.pet

915resolution supports GM45

Continuing from my blog post:

http://bkhome.org/blog/?viewDetailed=01251

Tempestuous has come to the rescue and added support for Intel GM45 video hardware. Forum thread, with latest source and a PET:

http://murga-linux.com/puppy/viewtopic.php?t=32462

Puppy Stardust

Those who love a slim-trim-and-elegant Puppy will like what zigbert has done. Zigbert has taken a standard Puppy 431 and tweaked it:

http://murga-linux.com/puppy/viewtopic.php?t=49971

The iso is 98MB.

Woof uploaded

I have uploaded Woof, which has the 'CHANGED_ONLY' option for '2createpackages' and GRUB4DOS support in the Universal Installer.

Note: I have also fixed ext4 support in the Quirky build. Strangely, the e2fsprogs package compiled in T2 does not support ext4, so I change over to our PET package.

2createpackages improved

There was a comment on the forum that it is a limitation of '2createpackages' in Woof that it can either build one package or build all of them -- the latter takes a long time.

If you have changed a few packages, say new version and/or added new packages, but can't recall which ones exactly, I have added a new feature to '2createpackages' that takes care of this.

In the GUI there is now a 'CHANGED_ONLY' selection, or it can be done from the commandline:

> ./2createpackages CHANGED_ONLY

This looks to see what packages have been changed/added since last time 2createpackages was run, and only rebuilds those.

I am just about to test it. If it works, I'll probably upload the latest Woof in the morning (which also has the GRUB4DOS-compatible Universal Installer).

Universal Installer improved

I reported earlier today about shinobar's GRUB4DOS package:

http://bkhome.org/blog/?viewDetailed=01252

I have made this the default in Quirky, removed GRUB.

I have modified the Universal Installer, /usr/sbin/puppyinstaller, in Woof. It now supports both GRUB and GRUB4DOS, and I made a couple of simplifications as well.

Bones bugfix

Thanks to playdayz report in the previous blog post, http://bkhome.org/blog/?viewDetailed=01249, I have fixed a bug. Get 'bones-11dec2009.gz' from here:

http://puppylinux.com/test/bones/

Playdayz, install this latest 'bones' script, then run 'bones download' again -- this time there should not be any error message.

For extra peace-of-mind, you can check afterward, run 'bones check', which checks that the database is correct.

GRUB4DOS

Forum member shinobar recently posted to the forum about the GRUB4DOS package used in the Japanese edition of Puppy:

Japanese Puppy: http://openlab.jp/puppylinux/index.html.en
GRUB4DOS PET package: http://shino.pos.to/linux/puppy/

I think that I read awhile back that the GRUB in pup 431 has a problem with number of inodes for ext4. Forum member CatDude has researched this and come up with a replacement, which also has a graphical boot interface:
http://www.murga-linux.com/puppy/viewtopic.php?t=49873

Both of these are very interesting. I decided to test shinobar's GRUB4DOS 0.4.4v1.4.1. It has a nice simple GUI and all seemed to go well, but when I booted all I got was this message:

GRUB Loading stage1.5

...then it hung.

The GRUB4DOS GUI offered to install to the MBR of sda, and I accepted the default. But, my PC has mixed SATA and IDE drives, and that is the problem. What Linux sees as sda (SATA), GRUB4DOS looks at the other one first (the IDE drive) -- and I previously had GRUB on the MBR of the IDE drive.

The solution was to choose sdb MBR to install GRUB4DOS.

I think that a solution to this problem would be for the GRUB4DOS installer to detect mixed IDE and SATA drives and default to install on the IDE drive. However, I don't know how to auto-detect that anymore -- on older kernels it was easy, as IDE drives had 'hd' naming, SATA drives were 'sd'.

I intend to modify the Puppy Universal Installer to work with shinobar's GRUB4DOS installer, so I could put in a warning.

I unplugged the IDE drive, then when I installed GRUB4DOS to sda, it booted ok. The problem is just that search-order, IDE found first.

One other point to note: I read somewhere that although GRUB4DOS will recognise ext4 partitions, it should not be installed to an ext4 partition. Which I take to mean, if installed to MBR of say sda, with 'menu.lst' and 'grldr' in sda1, then sda1 must not be ext4.

915resolution patched

'915resolution' patches the video BIOS of Intel video hardware to accept non-standard widescreen resolutions (not defined in the VESA standard).

A Quirky tester reported a problem with 915resolution on their video hardware, so I looked around to see if there are any recent patches.

Forum member tempestuous did all the work for us for Puppy 4.1.1:
http://www.murga-linux.com/puppy/viewtopic.php?t=32462

I am not aware of any Puppy developers patching it since then.

The Arch Linux guys have patched it fairly recently:
http://aur.archlinux.org/packages.php?ID=15486

The only extra one that Arch has added is 500GMA. One of the chips that tempestuous identified as 945GM, Arch has identified as 945GME, so I changed that. The only other small thing is that a chip that tempestuous named as GM965, Arch named as 965GM.

So, I have patched the source so that it now supports all of these:

"500GMA", "830", "845G", "855GM", "865G", "915G", "915GM", "945G", "945GM", "945GME", "946GZ", "G965", "Q965", "GM965", "G33"

The source is at:
http://bkhome.org/sources/alphabetical/9/915resolution-0.5.3-patched20091210.tar.gz

I have uploaded a PET package to here:
http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-quirky/915resolution-0.5.3-patched20091210.pet

Woof uploaded

I posted recently about improvements to placement of desktop drive icons on the desktop, and improved 'bones' script.

This is all in Woof and I have just uploaded it to the Bones online repository.

But, you do need the latest 'bones' script to download Woof, 'bones-10dec2009.gz', see previous post.

Intro and how to use Bones, see my blog posts:

http://bkhome.org/blog/?viewCat=Bones

Bones updated

I have improved the 'bones' script, and you now have to use this one. Get 'bones-10dec2009.gz' from here:

http://puppylinux.com/test/bones/

...gunzip it, set it executable, place at /usr/sbin.

The .delta files have a different naming, with both previous and latest dates in them.

There is improved database checking, and 'bones check' will perform a database integrity check.

If you have already downloaded Woof, running 'bones download' with the new script will automatically update to the new .delta files (I think).

I tested a brand new download, works fine.

Full control of desktop drive icon placement

In Puppy 4.3.x, the desktop drive icons are restricted to running along the bottom of the screen, just above the tray.

I have now made placement totally configurable. It is controlled by these variables in /etc/eventmanager:

#091208 adjust drive icons placement, read by /sbin/pup_event_frontend_d

#TODO: GUI in /usr/sbin/eventmanager

#gap between icons and edge of screen (>0 if need to leave space for a tray)...

ICON_PLACE_EDGE_GAP=64

#indent from edge before icons start...

ICON_PLACE_START_GAP=32

#spacing between successive icons (64 matches other desktop icons)...

ICON_PLACE_SPACING=64 #56

#left, right, bottom, or top of screen...

ICON_PLACE_ORIENTATION='bottom'

The script /sbin/pup_event_frontend_d reads these and draws the icons. Hopefully theywill avoid other icons.

Another improvement is that if there are too many partitions to fit on one line (or column), it wraps to another line (or column).

This is done in Woof, but I won't upload it just yet. I want to improve something in Bones first, so I'll upload in a day or two.

Anyone can put code into the 'pinstall.sh' script of a package to modify these variables. I have done so my 'rox-panel-quirky' package, to draw the drive icons right at the bottom of the screen. I also modified the default spacing from 64 to 56, as I think they look nicer slightly closer together.

Flwm, wmx window managers

Jemimah is using Flwm, so I took a look at it. First I compiled FLTK, version 1.1.10rc3:

# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --disable-gl --enable-xdbe --enable-xft --enable-threads --enable-xinerama --enable-largefile

Then Flwm (which needs FLTK):
I edited config.h:

#define CLICK_TO_TYPE 1
#define TITLE_FONT "DejaVu Sans"
#define MENU_FONT "DejaVu Sans"

# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --host=i486-t2-linux-gnu

...there were some compile errors, had to do a couple of hacks to get it to compile. Fltk was linked statically and I got a 'flwm' executable at 202KB.

I also got Flwm out of CVS, noticed the code is quite old, like 2000. It appears to be based on an old version of wmx. I thought it would support Xft (as wmx does) which is why I specified 'DejaVu Sans' font, but it only displays a bitmap font.

wmx version 7 is more recent code and does support Xft. It does not use FLTK. The executable is 140KB. I'm running it now, but it lacks the extra title-bar buttons that the Flwm developer put in, which makes it awkward to use -- like, how to maximise a window? -- you need to know a key-combination, can't do it with the mouse.

One interesting feature of both of these is they have the title bar on the side of the window, which might be good for small widescreen netbooks.

Flwm: http://flwm.sourceforge.net/
wmx: http://xwinman.org/wm2.php

Fluppy puplet

Jemimah is doing great work, optimising Puppy for small widescreens:

http://www.murga-linux.com/puppy/viewtopic.php?t=47192

"Fluppy" is a proposed puplet that will incorporate these ideas.

Password protected sources

I have added password protection to access:

http://bkhome.org/sources/

My laptop was chugging away all night uploading most of the source packages that I used for the T2 build.

The problem is though, when it becomes widely known that I have a very nice repository of source packages, and especially when people decide to download everything, I am going to run into problems with my host. My account with Hostgator has "unlimited downloads" however it isn't really, as it is one of those cheap shared accounts ("Baby Croc").

This "download everything" problem is going to occur with Woof Tpup builds, as I am uploading most of the packages that I used to build my latest Tpup. It is better that when people use T2, just let it try and get the required packages from the T2 repositories or the author's sites. My site can be a last resort.

My username and password is the same that Eric is using, "pup�py" and "lin�ux". -- probably not necessary, but they are displayed here slightly obfuscated.

Playing with Echinus and Oroborus

These are tiny window managers. Unfortunately they both have something wrong with them.

I'm running Oroborus right now, version 2.0.18. Looks nice. The only problem that I have encountered so far is that when I click an icon in the ROX Panel, nothing appears on the desktop -- I have to then click in the desktop for the application to appear. Weird.

Home: http://www.oroborus.org/

An aesthetic thing: these days window managers should support xft, even minimalistic ones.

Echinus has been discussed recently on the forum:

http://www.murga-linux.com/puppy/viewtopic.php?t=47476

I compiled version 0.3.9. It is delightfully tiny and simple. Unfortunately, when I right-click on a file in a ROX-Filer window the menu appears underneath the window! (only in floating mode, it's ok in tiled mode). Also, there doesn't seem to be any way to drag-resize a window.

So for me the only windows managers that I have been satisfied with are still JWM and Openbox.

Woof page updated

I have updated the Woof introduction page:

http://bkhome.org/woof/

This includes explanation of how to get Woof out of the Bones online repository.

Note, I have just now updated Woof online, so it's my very latest, that I used to build Quirky 001 -- plus I fixed a couple of bugs in the 'devx' (a couple of missing symlinks).

A warning about 'bones download'...
The first time that you run 'bones download' that's fine, you will then have Woof. After that, if you see an announcement on my blog that I have uploaded a new version of Woof, you can run 'bones download' again to sync with the latest code. Note however, if you have modified any of the files, they will get over-written.

You would need to backup any customised files first. This is an area that I plan to improve.

Note that if you have user customisations that you think I might want to incorporate into Woof, if you run 'woof save' it will create a delta file in directory 'woof-bones' that you can email to me.

Quirky version 0.0.1

The live-CD ISO file is 90.36MB and available from here:

http://distro.ibiblio.org/pub/linux/distributions/quirky/quirky-001/

There is also a README file:

http://distro.ibiblio.org/pub/linux/distributions/quirky/quirky-001/001-readme.htm

...which explains that this release only has a "quirkiness factor" of one.

I have only just uploaded it, it hasn't got through to the mirrors yet.

A clean screen for netbooks

I am working on a different desktop for Quirky, that is more usable for small widescreen netbooks.

A tray along the bottom or top (or even worse, both) is not the best for a small wide-screen, as there tends to be more "free space" not required by windows at the sides.

Then of course we have icons on the desktop, and windows have to be dragged aside to get at them.

I was inspired by Jemimah to take another look at using the ROX panel instead of the JWM tray. Right now I'm playing with a panel on the right side, the drive icons moved down slightly so they are right at the bottom of the screen, and all the other desktop icons removed.
...I'll probably upload the first Quirky ISO with this.

For the record, the packages required are:

libwnck-2.26.2-p4.pet
rox-clib-2.1.10-q1.pet
rox-clib_DEV-2.1.10-q1.pet
rox-menu-0.5.0.pet
rox-monitor-applets-1-p4.pet
roxpanel-pager-1.2-p4.pet
rox-panel-quirky-1.pet
roxpanel-systray-0.3.2-p4.pet
roxpanel-taskbar-0.4-p4.pet

These will be available on my ibiblio Quirky site.

'rox-panel-quirky-1.pet' has configuration stuff to set things up properly for Quirky.

Most of the other packages are just Jemimah's packages re-badged.

I don't know if I'll stay with this though, due to the size overhead. Next, I might experiment with a JWM tray on the right side. We'll see...

Why you should NOT visit Australia

Heh heh, see the forum:

http://murga-linux.com/puppy/viewtopic.php?p=369573#369573

Quirky page expanded

Here it is:

http://bkhome.org/quirky/

There is still no actual ISO to download, but I'm getting there. I keep finding stuff to either fix or improve.

Inkscapelite

This is also known as "InkLite" in the menu. Inkscapelite is a fork of Inkscape 0.36pre0, the last version to be written in C. After that, they rewrote it in C++ and also used GTKmm -- and my oh my did it ever jump up in size, especially if you linked the GTKmm libs statically.

Technosaurus and ttuuxxx were discussing Inkscapelite recently on the forum, in regard to its dependence on 'libgnomeprint' and 'libgnomeprintui'.

I have also taken those two libraries out of Quirky, so I wanted to recompile Inkscapelite either statically with those libs, or without -- Inkscapelite has a configure option '--without-gnome-print' which I chose -- the application can still print, but it is very basic.

The problem is, the only source package I can find is 'inkscapelite-0.36.2.tar.gz' but Puppy currently has version 0.36.3. Maybe I have it archived somewhere, but anyway I used the 0.36.2 source and compiled that -- one file needed a small hack to get it to compile, so perhaps that is what was done before when I created 0.36.3. So I decided that I might as well name this as version 0.36.3 and I have uploaded it here:

http://bkhome.org/sources/alphabetical/i/

If anyone has that original 0.36.3 source, I would like to see it, find out if I did anything else.

T2 web page

I have updated my web page on T2:

http://bkhome.org/t2/

gcc 4.3.4 recompiled

An old problem has come back. Previously when I compiled in T2, such as the base packages for Puppy 4.0, there was a problem with the gcc compiler. When using it in an alpha of Puppy 4.0, some packages would not compile, with error "__sync_fetch_and_add_4 undefined". I reported this, and the solution, in my news archive (do a search for "gcc bug fixed"):

http://www.puppylinux.com/news/news400a3-400a5.htm

With the recent T2 compile I made some changes that I thought would have fixed this problem, but not so. It has to do with the cross-compile environment that T2 runs in. So, here we go again...

These are patches I got out of T2, and the configure parameters are based on information in var/adm/logs/5-gcc.log in the T2 build:

# patch -p1 < ../fixincl.patch
# patch -p1 < ../fixincl2.patch
# patch -p1 < ../libstdcpp-with-tag-cc.patch
# patch -p1 < ../no-install-libiberty.patch
# patch -p1 < ../stdc++-include_next.patch

# mkdir objdir
# cd objdir
# ../configure --prefix=/usr --bindir=/usr/bin --sbindir=/usr/sbin --libdir=/usr/lib --datadir=/usr/share --includedir=/usr/include --infodir=/usr/info --mandir=/usr/man --sysconfdir=/etc --localstatedir=/var --disable-debug --without-libpam --without-pam --disable-libpam --disable-pam --build=i486-t2-linux-gnu --enable-__cxa_atexit --disable-checking --disable-bootstrap --disable-libstdcxx-pch --disable-multilib --with-gnu-as --with-gnu-ld --enable-threads=posix --disable-libgcj --enable-languages=c,c++,fortran --enable-shared

...it's compiling right now. Gonna take awhile, it's a big package.

Note, this time I have enabled the Fortran compiler, which is missing from the Puppy 4.x series.

This is delaying the upload of the first pre-alpha of Quirky for a couple more days. I found bugs with Abiword 2.8.1, so I wanted to try compiling Abiword from SVN, and I also recompiled the Enchant package or rather tried to -- I hit that "__sync_fetch_and_add_4" bug.

2.6.31.5 kernel

I am using the 2.6.31.5 kernel for Quirky.

The patched source tarball and SFS are here:

http://bkhome.org/sources/kernel-2.6.31.5/

Logos from afgs

Forum member afgs has been doing great things with the logos:

http://murga-linux.com/puppy/viewtopic.php?t=48880

Here is a Quirky logo:

And this is what it looks like in a web page:
http://bkhome.org/quirky/

Here is a Woof logo:

And this is what it looks like in a web page:
http://bkhome.org/woof/

SM2 rendering problem

I have just booked a return flight to Melbourne to visit my daughter next year. Tiger Airways are offering advance booking low fares ($100 each way).

Before booking, I tried to look at the prices on offer from Jetstar:

http://www.jetstar.com/au/en/index.aspx

...however, rendering is awful. So bad I was unable to fill in the flight search form.

I'm using SeaMonkey 2.0 every day now, and when I first started using it I was pleased that now I am using the same rendering engine as Firefox 3.5. But now I'm not so pleased.

Anyone reading this who is running SM 1.1.18, does that site render ok for you? Firefox? Opera?

I'm increasingly thinking I might go back to SM 1.1.18.

Another thing that is broken in SM2 is drag-and-drop of URLs. I used to be able to drag a file from the web browser window to any open ROX-Filer window, and wget would start and download it -- very convenient -- wget is much better than the SM Downloader. However with SM2, wget reports that the URL is broken.

Testing Bones

Seems to be ok for basic usage. I have uploaded the Woof project as a Bones repository. So, to get hold of the latest Woof, do this:

1. Obtain latest 'bones' script
From Here: http://puppylinux.com/test/bones/
...gunzip it, set it executable and place at /usr/sbin.

2. Bones setup
In a Linux partition, do this:
> mkdir woof-tree
> cd woof-tree
> bones setup
...enter a local username
...enter url for woof project: http://bkhome.org/bones/woof/

3. Download Woof
> bones download

After step 3 you are ready to use Woof.

Further information: http://bkhome.org/bones/
Please report any bugs!

More Bones

I'm back! I was in Perth for the weekend + Monday.

I did a little bit of work on Bones, fleshed out the script a lot. I focused on getting the round-trip to work, that is, from the administrator (me) saving, uploading, then the user downloading, using, maybe contributing to Woof development.

Well, I did what I could without having an actual Internet connection. My daughter on the otherhand, had an iPhone plugged into her laptop and had Internet access.

Anyway, I wrote-up progress so far:

http://bkhome.org/bones/

Bones version control

A personal, tiny, and very simple version control system that does what I want for managing the Woof project:

http://bkhome.org/bones/

Offline for a few days

An announcement to let everyone know why I won't be posting here for awhile.

My daughter is flying over from Melbourne to Perth for the weekend, so I'll be in Perth staying with one of my sisters -- she doesn't have an Internet connection.

However, I will be continuing to work on the Tpup build, and plan to upload an alpha when I get back home. It works fine right now, I'm very pleased and using it as my regular daily distro. But, I should do more testing and refining.

There's a difference between Tpup and Quirky, in case you're confused! Tpup is just your usual Puppy-derivative or puplet built with Woof, similar to Dpup and Upup but instead built from T2 packages.

Quirky just twists it a bit, adds the quirks. In Woof, it's an extra script, '4quirkybuild'. The only quirky thing that I have done so far is put everything into the 'vmlinuz' kernel so that the entire distro is just one file.

So I'll describe the build as "Tpup" if it is a normal puplet, and as "Quirky" if it has the quirks. Early next week I might just upload Tpup, or I suppose I could upload both.

There's another quirky idea I'm keen to work on, soon as I get the time. The days aren't long enough for me!

Printing now works

Testing my Tpup, I recently reported trouble with printing. As reported, I rolled back to CUPS 1.3.11 which fixed recognition of my USB printer, however I got a "foomatic-rip failed" message when I tried to print.

It turned out to be a missing Perl script that foomatic-rip requires, 'Hash.pl'. This script is part of the Perl package, however Puppy uses a cutdown Perl, very cutdown. I modified the Perl template in Woof to include this script.

Then I discovered a problem in /usr/sbin/cups_shell. As I'm using SM2 I don't have MU's lovely PuppyBrowser and have to use SM as the internal help viewer, also for the CUPS web interface.
cups_shell starts SM with a help page, then tries to open the CUPS web interface. The latter fails as SM needs a couple of seconds to start the help page, more time on older PCs. I fixed this.

Hey, my latest Tpup is just 90MB!

SeaMonkey 2.0 quibbles

For day-to-day usability I'm not so happy with SM 2.0 compared with 1.1.18. Three quibbles:

1. Keeps asking if want to remember password, even when I reply "never for this site".
2. Sometimes at startup, SM apologises that previous session did not close down properly and do I want to restart with previous session or a new session. However, I did close SM down properly previously.
3. SM remembers form and search entries. Trouble is, when I login to gmail, I type my username and SM displays it in a drop-down list which covers the password field. The drop-down won't go away unless I click elsewhere. Annoying. I changed Preferences to not remember form and search history.

So, I don't think SM 2.0 is quite ready for prime-time.

Dpup beta5 coming

Gposil and team have been busy developing the next-generation Puppy, that we have been terming "Puppy5". Gposil has announced that beta5 should be uploaded tonight. Follow development here:

http://murga-linux.com/puppy/viewtopic.php?t=47592

Here is the Dpup home site:

http://www.dpup.org/

I was just thinking yesterday, we have the "pet_packages-5" directory sitting virtually empty at ibiblio.org. So, that could be a home for Dpup PET packages. I won't manage uploading to it though -- Gposil or someone can have ftp access to it.

Another Woof logo

My daughter has designed a Woof logo:

...I messed it up a bit though. Making it's proportions square, some shading is chopped off the bottom. if anyone wants to tweak it, go for it.

I also put it into my web page to see what it looks like:

http://puppylinux.com/woof/

...quite nice!

Xorg Wizard bugfix

I'm knocking off the bugs with this T2-based-pup.

I'm running Xorg with the 'vesa' driver, and Xorg Wizard does not display the vertical and horizontal frequencies when testing a video mode. Fixed.

T2 binary packages

The third run of T2 was completed yesterday, and I am pleased with the outcome. So, I have once again uploaded all the binary packages that were compiled in T2 to here:

http://bkhome.org/binaries/t2/

These are the raw packages that can used by Woof to build a "tpup". I plan to use these as the basis for Quirky Linux.

Eliminate Type1 and most bitmap fonts?

I think that ttuuxxx and kirk have both been down this track. That is, replace most of the Type1 and bitmap fonts with TTF fonts. Objectives: better looking and save space.

I have just done a bit of a search on the forum and can't find this documented. Guys, have you posted anything about this anywhere?
I have a vague memory that ttuuxx sent me something awhile back.

The T2 build has Xorg 7.5, which is working fine. Now, 7.5 is much more TTF-capable, so opportunity is definitely there to remove the older fonts.

The only reservation I have is Ghostscript, that may have some need for Type1 fonts to render some .ps and .eps files.

MediaPlayerConnectivity

This looks good. URL:

https://addons.mozilla.org/en-US/firefox/addon/446

It doesn't work in SeaMonkey 2.0, so I have sent an email to the author asking if it can be adapted to SM2.

Testing T2 build

I did a recompile in T2. This is the third one recently. It took about 30 hours.

I'm aiming for small size, so am making quite a few radical decisions. The live-CD is 92MB, although I am aiming much lower than that.

Working nicely. Some issues of course. The compiling environment works -- I compiled Gxine.

And I'm online running it right now, using SeaMonkey 2.0.

Next thing to do is test printing....

I mentioned before, I'm doing this mainly for my Quirky Linux, which has one objective to be very small. Though, it's fine for any puplet. When I fix a few more bugs, I'll upload another Woof tarball and also the T2 .tar.bz2 binary packages (the Woof build system can download these to build a live-CD distro).

Keyboard layout

Continuing from the previous post, I have revised how keyboard layout is handled.

The modified scripts are 'init', 'rc.country' and 'input-wizard'.

The relevant data files are no longer in /usr/share/kbd, instead are in /lib/keymaps and /lib/consolefonts. Both of these are in the initrd, and the 'pkeys' boot parameter can now specify any layout and it will be loaded by the 'init' script.

'pkeys' boot parameter

In Puppy 4.3.1, the 'pkeys' boot parameter can be used to change the codepage and iocharset when mounting a vfat filesystem.

The pkeys boot parameter is also read in /etc/rc.d/rc.country after switching to the main Puppy filesystem, and if set then the appropriate keyboard layout is loaded (and on first boot the dialog asking for keyboard layout will not appear).

A limitation though, pkeys is currently restricted to:
be br cf de dk es fi fr gr hu it jp no pl ru se uk us

These are the keymaps located at /lib/keymaps.

There are a number of improvements that can be made, but for now I have implemented one:

The keymap is loaded after the switch-root to the main Puppy filesystem. The problem is, if you have to respond to any questions in the initrd, such as to choose a 'pupsave' file from multiple choices, the keyboard is at the default layout (us).

Woof now places /lib/keymaps into the initrd, and if there is a 'pkeys' boot parameter, for example 'pkeys=uk', then that layout will get loaded by the init script (that's one reason why I recompiled busybox, to get the 'loadkmap' applet).
The init script copies /lib/keymaps to the main filesystem before the switch-root.

Future improvements... I'm thinking about it. Forum member ecube did some work excellent on this back in April 2009:

http://murga-linux.com/puppy/viewtopic.php?t=40658

Ecube also sent me a pm with modified Keyboard Wizard and rc.country. Ecube, thanks for that -- I sure have taken awhile to get around to looking into it!

The Tiny Core Linux people converted the kbd-1.14.1 package to kmap format that Busybox's 'loadkmap' applet can load. The kmap files that I currently have in /lib/keymaps are very old, go right back to the very first release of Puppy in June 2003.

Woof tarballs

As I have put Fossil on hold, and quite likely won't reinstate it, for now I am uploading tarballs of Woof to here:

http://bkhome.org/binaries/woof/

Regarding a suitable version control system, I'm thinking that I will work with the Dim developers, as I think that is a much more Unix/Linux friendly system. See my earlier blog report:

http://bkhome.org/blog/?viewDetailed=01191

...Dim currently does not support symlinks, empty directories or special files either, but the developers are intending to support those, and I might get involved.

Developing Dim though, is a longer term thing, so I'll stay with uploading Woof tarballs for awhile.
They are only 8MB so it's easy enough to do frequent uploads.

Fossil on hold

C**p, problems with Fossil.

When I first created the fossil repository for Woof, it did store symlinks. However it has now converted them all into files.

Woof has so many symlinks.

So, Fossil is on hold.

Busybox recompiled for initramfs

The initramfs ("initial ramdisk") has busybox version 1.4.2, statically compiled with uClibc. That was last done on 12 Nov. 2008. I have now recompiled it, kept the same version, and added more applets, plus enhanced the features of some. Fascinating, the busybox executable has grown from 253KB to 272KB, only 19KB, even though I have turned on help-text for each applet (the help text is internally compressed).

I have reasons for wanting these extra applets, as will come clear in time. Here is the Fossil commit:

http://bkhome.org/fossil/woof.cgi/vinfo/8fbef54932b0355f6d3bb62fe6a98ecf4a56e875

I turned on support for '--help', so am able to provide the following information:

NEW APPLETS

dc

Usage: dc expression ...



This is a Tiny RPN calculator that understands the

following operations: +, add, -, sub, *, mul, /, div, %, mod, **, exp, and, or, not, eor.

For example: 'dc 2 2 add' -> 4, and 'dc 8 8 \* 2 2 + /' -> 16.



Options:

p - Prints the value on the top of the stack, without altering the stack

f - Prints the entire contents of the stack without altering anything

o - Pops the value off the top of the stack and uses it to set the output radix

    Only 10 and 16 are supported

    

halt

Usage: halt [-d] [-n] [-f]



Halt the system



Options:

        -d      Delay interval for halting

        -n      No call to sync()

        -f      Force halt (don't go through init)



loadfont

Usage: loadfont < font



Load a console font from standard input



loadkmap

Usage: loadkmap < keymap



Load a binary keyboard translation table from standard input



more

Usage: more [FILE ...]



View FILE or standard input one screenful at a time



pipe_progress

No help available



poweroff

Usage: poweroff [-d] [-n] [-f]



Halt and shut off power



Options:

        -d      Delay interval for halting

        -n      No call to sync()

        -f      Force power off (don't go through init)



reboot

Usage: reboot [-d] [-n] [-f]



Reboot the system



Options:

        -d      Delay interval for rebooting

        -n      No call to sync()

        -f      Force reboot (don't go through init)



setkeycodes

Usage: setkeycodes SCANCODE KEYCODE ...



Set entries into the kernel's scancode-to-keycode map,

allowing unusual keyboards to generate usable keycodes.



SCANCODE may be either xx or e0xx (hexadecimal),

and KEYCODE is given in decimal



setlogcons

Usage: setlogcons N



Redirect the kernel output to console N (0 for current)

ENHANCED APPLETS
The following applets existed before, but I have enhanced their features...

pidof

Usage: pidof process-name [OPTION] [process-name ...]



List the PIDs of all processes with names that match the

names on the command line

Options:

        -s      Display only a single PID

        -o      Omit given pid

                Use %PPID to omit the parent pid of pidof itself



readlink

Usage: readlink [-f] FILE



Display the value of a symbolic link



Options:

        -f      Canonicalize by following all symlinks



tr

Usage: tr [-cds] STRING1 [STRING2]



Translate, squeeze, and/or delete characters from

standard input, writing to standard output



Options:

        -c      Take complement of STRING1

        -d      Delete input characters coded STRING1

        -s      Squeeze multiple output characters of STRING2 into one character



...it is not documented here, but enhanced functionality is to convert between upper-case and lower-case.

Smaller Xorg

I have been working on making Xorg smaller in Woof builds.

I have removed the "openGL" (and DRI drivers) component by default in Woof builds, as with Xorg 7.4 and 7.5 that has become enormous. The template was cutting it down a lot any way, now I have taken it right out. It can be added back in as a separate package later on.

I am mostly aiming the T2 build for Quirky, where I want very small size. I am experimenting with cutting Xorg down to the 'vesa' driver only (not the kdrive Xvesa, the vesa driver that is part of Xorg).

The uncompressed size of /usr/X11R7 is as follows, for my T2 build:

Before: 15.3MB
GL removed: 12.58MB
Vesa driver only: 8.02MB

Fossil commits:
http://bkhome.org/fossil/woof.cgi/vinfo/04c53c08ba048692414d78b749d6c215f7a1db5d
http://bkhome.org/fossil/woof.cgi/vinfo/be5207f38d91e576c2fa226564ff5f16a6aff2e9

Xorg Wizard multiple monitors

Pizzasgood posted a patch to the 'xorgwizard' script in the pup 4.3.1 forum feedback thread.

I have applied this to Woof:

http://bkhome.org/fossil/woof.cgi/vinfo/fec6837bd2bfca364084f05e8329576459f7ed46

PDQ print system

Experiencing problems with CUPS again, I remember wistfully the days when Puppy had the PDQ print system. I posted a comment to the previous post, but thought this deserves it's own post.

PDQ homepage:
http://pdq.sourceforge.net/

Early versions of Puppy had PDQ, which is very nice, really tiny. It fell by the wayside for various reasons:

No longer being developed.
GTK 1.2 GUI application only.
Some apps expect CUPS to be there.
Very basic printer configuration.

We could save an incredible amount of space with PDQ, plus it is very simple, and "just works". But, it would need someone who is able to commit to developing it. It is written in C. For a start, the GUI application would have to be rewritten for GTK2.

Mandrake Linux used to use PDQ as the default printing system. Um, I can't recall what version was the last to do that... 8.x comes to mind. The thing is, those guys had developed PDQ to a high art, did very nice things with it. But, like all the other distros they dropped it and went for CUPS

So, if anyone wanted to take on PDQ as a project, getting an old Mandrake distro would be useful.
I don't recall the last version of Puppy that used PDQ -- it would be in my online blog archive.

I had developed a setup GUI that asked various questions about your printer, like how many color nozzles in an inkjet cartridge, so it didn't need a big database of printer types, but that would probably need to be developed further -- which is where the Mandrake stuff would be useful.

I wouldn't mind messing around with it again. Anyone interested in helping out with porting it to GTK2?

CUPS 1.4.1 does not work

My T2 build has CUPS 1.4.1, but it does not recognise my USB printer.

I started earlier today to trace reasons why Woof builds are bigger than earlier puppies. I'm looking at the T2 build for now. The Ghostscript package for example is massive compared to the one in pup 4.3.1.

This lead me to testing printing, when I discovered CUPS won't recognise my printer. It doesn't recognise its existence. A quick search revealed this is a known problem:

http://bugs.archlinux.org/task/15998

The kernel 'usblp' module is no longer used, so I blcklisted it. But that did not help. The above link advises further changes, so I need to investigate further.

Sigh, CUPS is such a pain.

Bug trackers!

I have activated the bug tracker for Woof, and posted a test-ticket:

http://bkhome.org/fossil/woof.cgi/rptview?rn=1

To view my ticket, to create a new ticket (bug report) and to append comments to existing tickets, you have to be logged in as 'anonymous'.

The requirement to login is partly a security measure and partly to prevent spambots from placing extra drain on the server.

To avoid confusion, I think it best if only users of Woof report bugs to my ticketing system. That is, guys using Woof to build a distro, for example ttuuxx, gposil and kirk.

For all Puppy Linux bugs, technosaurus has setup this bug tracker:

http://code.google.com/p/puppy-development/issues/list

However, regardless of what version of Puppy you are using, if you know there is a problem in a particular script that is in Woof, and you have looked at the Fossil version control system for Woof:
http://bkhome.org/fossil/woof.cgi
and seen that the latest script has the same problem as the script that you are using, then certainly report the bug to my Woof bug tracker.

...or even fix it and send me a patch!

POST EDITED NOV 19

Weird shutdown problem

I was testing a T2 build, and when I tried to do a shutdown, it asked if I wanted to save the session, I declined, then X restarted!

Hmmm. There was something different from normal. Usually I unmount all partitions before closing down, but this time I had left one mounted.

The /etc/rc.d/rc.shutdown script attempts to kill any processes running on that partition (then later unmount it). take the example of /dev/sdc3 mounted, this is executed:

fuser -k -m /dev/sdc3

However, there are no processes running on /dev/sdc3, as 'fuser -m /dev/sdc3' confirms.
However that line causes a weird event to occur -- X immediately restarts!

I have put a workaround into rc.shutdown:

http://bkhome.org/fossil/woof.cgi/timeline

However, I really do need to investigate this some more.

Note, you will need to login as 'anonymous' to see the actual changes made to rc.shutdown.

Fossil: Limitation workarounds

I have identified four limitations of Fossil that causes problems with hosting the Woof project. As far as I know, these limitations also exist in GIT and SVN. I have added to my Fossil Quick Guide page explaining these limitations (scroll down near bottom):

http://bkhome.org/fossil/fossil-guide.htm

I developed a wrapper script named 'fossil' to workaround these limitations (and I renamed the binary executable to 'fossil.bin'). This script (and fossil.bin) is in a PET package, linked-to in the Fossil Quick Guide.

Direct link to PET package:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/pet_packages-4/fossil-20091111.pet

Note: I am currently doing some testing that these Fossil workarounds do actually work.
Then I'll get back to fixing some bugs in the T2 build (Tpup).

Fossil: Woof repository not ready

Just in case you go to my new Fossil version control repository and see all files are uploaded:

http://bkhome.org/fossil/woof.cgi/timeline

Don't be tempted to do a checkout yet, as described here:

http://bkhome.org/fossil/fossil-guide.htm

...because it won't work. That is, the Woof build scripts will fail.

Fossil cannot store empty directories, files with '[' or ']' in them, and device nodes. Therefore I have had to employ workarounds to get them in.

The build scripts will have to be aware of these workarounds, and I have not yet done that. Hopefully will do that tonight or tomorrow.

As far the the T2 build, there are still some bugs that need to be fixed, plus I will have to upload the latest set of raw binary compiled packages.

I'll announce when it is ready.

Note, once it's up, I'll be syncing with it every time I make a local change, so you will be getting my very latest Woof.

Fossil: database integrity

I have a very temperamental Internet connection -- dish on the roof pointing at a satellite. Apart from the over-one-second round trip, the connection drops out randomly then comes good again seconds or minutes later -- clouds? atmospheric affects?

Anyway, one of these dropouts occurred while I was syncing my local Fossil repository with the online one, which caused the operation to hang. It then raised the whole question about possible corruption of a repository.

I raised this question in the Fossil mail list:

http://www.mail-archive.com/fossil-users@lists.fossil-scm.org/msg00833.html

Dr Hipp also referred me to this page:

http://www.fossil-scm.org/fossil/doc/tip/www/selfcheck.wiki

The precautions taken against any kind of corruption are very impressive.

Note, I have updated this page, added a Fossil Quick Guide, but it needs more work:

http://bkhome.org/fossil/

Woof: SeaMonkey 2.0

I am currently refining Woof to build a distro from packages compiled in T2.

The T2 build is pretty up-to-date with the latest packages, including SeaMonkey 2.0.

Woof has a directory 'packages-templates' which contains "templates" that define how a raw binary package is to be cutdown and modified to be Puppy-compatible. I have updated directory 'packages-templates/seamonkey' to handle SeaMonkey 2.x.

Actually, it isn't easy. It has to handle a directory layout that has been butchered by Debian/Ubuntu as well as normal layouts. The raw binary package compiled in t2 is of course "normal".

Apart from cutting it down a bit, there are little things to tweak. For example the 'prefs.js' file has to be modified so that the first time SM is run the Puppy home page comes up and not the SM home page.

Barking dog wanted

Has anyone seen a picture of a barking dog online, that I could use for the Woof logo?

I'm currently using this, as a temporary placeholder:

I got that off the Internet somewhere and it may have license restrictions, like for personal use only.

Ideally I would like a circular logo, to match other logos on my website (and the Puppy logo).

I'm still experimenting with the Fossil version control system. The more I learn, the more I like it. My experimental setup has moved slightly:

http://bkhome.org/fossil/

...I'm going to start experimenting with putting some files into it, and generally play around. If my confidence reaches a certain level, then I'll "go for it" (Fossil as the official Woof version control, wiki and bug tracking system).

Zombies on TV

Until a few days ago, I could only receive one TV channel, our ABC (Australian Broadcasting Commission) Government channel.

Now, with satellite TV, all sorts of stuff. I am only getting free-to-air channels. Well, kind of -- most of them are scrambled and require an Optus smartcard to view, a one-off purchase ($85).

I now get two commercial channels. I must say, after not watching any commercial TV for 10 years, my impression is these are mostly junk.
Many of the programs are lowest-common-denominator (sex, violence, money).

I now get SBS, which is a Government ethnic channel, sometimes has interesting stuff. On Monday night I watched "Dead Set", first of a 5-part series:

http://en.wikipedia.org/wiki/Dead_Set_(TV_series)

After that, I turned to another channel, and lo-and-behold, yet another zombie program, "Undead":

http://www.undeadthemovie.com/

I missed the beginnings of both of these, so don't know exactly how these zombies got going. It seems, in the Undead movie it is a rain of meteorites that infect the people. Zombies attack others, if bitten you become a zombie also.

Oh dear, hasn't this been "done to death" already?
I'm not into horror movies. Though, this undead thing is kind of amusing because it is so dumb.

Pwireless2

This is extremely interesting! Forum member jemimah has rewritten our Pwireless program (developed by plinej (Jason)):

http://murga-linux.com/puppy/viewtopic.php?t=48816

Quoting:

As some of you may know, wpa_supplicant is a network management daemon that runs in the background on your computer and connects you to wireless networks. It supports open networks, WEP networks, and WPA networks.

Puppy supports wpa_supplicant with the Network Wizard, but not in roaming mode. Roaming mode is a godsend for people with laptops who may need to reconfigure their wireless network several times a day. In roaming mode, you set up each network one time, and wpa_supplicant will automatically connect to it whenever you're in range. If multiple networks are in range, you can set priorities to tell wpa_supplicant which one you prefer.

Until now, the only way to configure wpa_supplicant roaming mode was by editing the configuration files by hand, or use wpa_gui (http://www.murga-linux.com/puppy/viewtopic.php?t=47838). Wpa_gui has a somewhat confusing interface and requires large Qt libraries, so I thought it was time someone wrote a lightweight GUI frontend that was specifically designed for and integrated into Puppy.

Pwireless2 looks much the same as the original Pwireless but the guts are totally replaced. Pwireless2 uses wpa_cli to communicate with wpa_supplicant. Wpa_supplicant handles profile storage, and all network configuration tasks. I've also included an upgraded version of Dhcpcd which handles allocating Ip addresses and notifying you when the status of your network changes.

Finally, I've modified a program called gtrayicon to provide you with a very basic networking tray applet. Clicking on the applet will tell wpa_supplicant to start or stop (clicking it if you haven't configured any networks won't do much of anything). Right click and choose Configure to launch the Pwireless2 configuration gui.

Jemimah, this is a great initiative.
Guys, test it!

A note on the original Pwireless. Plinej created it, and HairyWill and urban-soul did some further work. I think the most recent version is 0.8.4:
http://www.murga-linux.com/puppy/viewtopic.php?t=23121

Fossil progress

I have made progress, of sorts. I logged into my Fossil Woof online repository and changed permission to 'clone' from 'anonymous' user to 'nobody'.

I then did this:

> fossil clone http://bkhome.org/fossil/woof/woof.cgi woof.fossil

...it has downloaded the correct 'woof.fossil', which has my modifications (title, logo).

Hmmm, the Quick Start guide said nothing about having to do that.

Note, it is essential to have 'woof.cgi' on the URL in the above clone operation.

I'll follow through the other things I did in the last post, see if it all works this time...

Fossil version control

I was recently evaluating Dim version control, see my blog report:

http://bkhome.org/blog/?viewDetailed=01191

Dim is a Bash/Ash shell script, and a PHP script for the web site. Very small, but at this stage there is a lot on the to-do list.

Then technosaurus informed me about Fossil, a "small and simple" version control system. So, I have now started to evaluate Fossil.

Fossil homepage: http://www.fossil-scm.org/

...read that, and it sure does seem impressive!

Fossil has a single binary executable, named 'fossil'. I compiled it and it is 652KB. That's the whole thing, all you need to install Fossil, and the executable even has the Sqlite database compiled statically into the executable -- which is good for uploading to a website that might not have shared Sqlite library (or wrong version).
However, technosaurus has figured out how to compile Fossil against shared Sqlite library if you want.

Install on website
Having had just about zero experience with installing and setting up a version control system (only a limited play with Dim), I was scratching my head. But forged ahead, and it turned out to be a simple procedure, very simple. It will be useful to others I reckon, for me to document it here...

My first problem was where to upload the 'fossil' executable, as I only have a shared hosting website (with hostgator.com). I can't just bung it into the system /bin directory! Well, I took a punt, and created a 'bin' directory: /home//bin.
/home/ is where I am when I do an ftp or ssh login.
I then uploaded 'fossil' to the 'bin' directory. I set permission of 'bin' to 755 and 'fossil' to 755.

I then created directories /fossil/woof, set permissions at 755.
Inside 'woof' I created a script 'woof.cgi' (755) with this in it:

#!/home//bin/fossil

repository: /fossil/woof/woof.fossil

...of course, if you want to do this yourself, you will have to substitute the appropriate text into "< ... >".

I also created a '.htaccess' file (644) in fossil/woof directory with this in it:

AddHandler cgi-script .cgi

Options ExecCGI FollowSymLinks

RewriteEngine On

RewriteBase /fossil/woof/

RewriteRule ^$ woof.cgi

...I wanted to completely hide the 'woof.cgi' script from the URL that users would see, but I couldn't figure out how to do that. Instead, the above just allows a URL "http://bkhome.org/fossil/woof/" which will run the 'woof.cgi' script -- but then the user still sees the full URL with "woof.cgi" in it. If anyone is an expert on Apache mod_rewrite, advice welcome!

Almost there. I needed an empty "repository" to get going. I created it like this (locally, on my own computer):

> fossil new woof.fossil

I then uploaded 'woof.fossil' to fossil/woof by sftp and set permissions to 744.

Finally, I pointed my web browser here:

http://bkhome.org/fossil/woof/

...hey man, it works! A very pleasant experience, I must say. I just followed my nose, doing logical things to set it up, not many steps required, and it works as advertised.

Install on local computer
Just copy the 'fossil' executable to the execute-path, such as /usr/bin.

As far as usage goes, well, I'm just getting started. As I commented above, I have zero experience. Yeah, I have had occasion to checkout from various CVS/SVN/GIT hosted projects, and always followed whatever instructions they provided. Beyond that, I know nothing.

I haven't yet put any files into my 'woof.fossil' repository. It is important to note that the entire project resides in that one file. It has its own internal Sqlite database, and all files, all revisions, will be inside 'woof.fossil'. So, the file may grow quite large!

An obvious first thing to do is, how to download my online 'woof.fossil' repository to my local computer?

A good place to look for basic usage instructions is the Fossil Quick Start Guide:
http://www.fossil-scm.org/index.html/doc/tip/www/quickstart.wiki

Alright, this seems to be what is needed:

> fossil clone http://bkhome.org/fossil/woof/ woof.fossil

...amazing, it works! Be sure to write down the username and password it reports.

I notice though, it did not download the configuration I had done to the online woof.fossil -- title and logo -- the downloaded woof.fossil is empty, no default title and logo -- so, is something wrong?

To use the web interface for looking inside the repository, run:

> fossil ui woof.fossil

...at the time of writing I have not yet configured 'woof.fossil' to auto-launch the web browser, so manually start your web browser and look at:

http://127.0.0.1:8080/

When finished, CTRL-C in the terminal window to kill the fossil web server.

The logical next thing to do is checkout a project-tree. That is, extract the latest version of the project from the repository:

> mkdir woof-tree
> cd woof-tree
> fossil open ../woof.fossil

Now I want to create a file in my project, and add it to the repository:

> echo "test file number one" > NewFile1
> fossil add NewFile1
> fossil commit -m "first test commit NewFile1" --nosign
> fossil close
> cd ..

Problems

1.
The downloaded repository does not have stuff in it that I had put into the online 'woof.fossil'.

2.
"Autosync" is by default turned on, and when I ran the above commit, it reported that it was also uploading to my online repository. However, when I point my browser at the online Woof repository, there is nothing, zilch.

3.
Unfortunately, it gets worse. When I attempted to reopen my local 'woof.fossil':

> cd woof-tree
> fossil open ../woof.fossil

Segmentation fault

...but it was then open.

So, the current situation for me is that it does not "just work". Probably I'm doing something wrong, as there are obviously other happy Fossil users. Anyway, I'll keep experimenting.

I looked at my website error log, and see this:

File does not exist: /fossil/woof/xfer

I did try various ways of uploading. For example, these two:
> fossil push http://bkhome.org/fossil/woof/woof.cgi --repository woof.fossil
> fossil push http://bkhome.org/fossil/woof/woof.cgi/xfer --repository woof.fossil
...in both cases got a "bad command".

Woof GUI improved

Previous templates
I have added a pull-down list of templates in the first tab of the Woof GUI:

The "Previous templates" pull-down list currently has these entries:

Puppy 4.3.1
Ubuntu Karmic Koala
T2 8.0rc

These are recent successful builds. Choosing one of these will configure the above entries in the GUI window so that Woof should build a live-CD without problems.

So for example, you choose "Ubuntu Karmic Koala", then all the above entries get preset to build a Upup from Karmic deb packages. But then you can modify the individual boxes, such as change the version number.

I have implemented this as I realised this first tab was a bit confusing.

Default tab
Thanks to Zigbert who reported how to set default tab in a gtkdialog tabbed interface:

http://www.murga-linux.com/puppy/viewtopic.php?t=38608

...I have implemented this in Woof.

puppylinux.org hacked into

For your information. I don't know if it can compromise your own system by visiting it, best to stay away.

I have received a few emails about it, but note that I don't have anything to do with maintaining puppylinux.org.

There's a forum thread discussing this:

http://murga-linux.com/puppy/viewtopic.php?t=48695

Preliminary testing Xorg 7.5

My T2 build has Xorg 7.5, at least a pre-release of 7.5.

I used Woof to build a distro, booted off a Flash drive, got a black screen, dead keyboard.

I tracked it down to this line in the Xorg Wizard:

Xorg -configure

By bypasing that, I was able to get Xorg to start.

By running:

Xorg -configure > /mnt/sda6/x.log 2>&1

I was able to get a log, and I found this:

(EE) LoadModule: Module i810 does not have a i810ModuleData data object.
(EE) Failed to load module "i810" (invalid module, 0)

then it complained that vmware_drv.so has an undefined symbol, then it hung.

Anyway, one thing at a time. I could only find two partially useful links about that particular i810 problem:

http://www.mail-archive.com/universe-bugs@lists.ubuntu.com/msg128432.html
https://bugzilla.redhat.com/show_bug.cgi?format=multiple&id=439045

...the first link "solves" the problem by hardwiring use of the 'intel' driver into Xorg.conf.

...the second solution is very frustrating, as the fellow says he solved the problem, but does not say how.

UPDATE
Here's another i810 link:
http://www.linuxquestions.org/questions/slackware-14/what-happened-to-xorg-in-slackware-13.0-intel-chipset-762618/

...interesting, Slackware 13.0 has an alternate (older) intel driver that people can use if the "proper" one doesn't work.

Anyway, I have to go back and check what kernel modules were loaded. It looks like the i810_drv.so driver needs kernel DRM module i810.ko (and drm.ko). Yeah, this is new stuff -- previously only the agpgart.ko and intel_agp.ko modules were needed. I'll check and report back.

Puppy 4.4 links

As it was asked about recently, here are the forum links that I know of, where Puppy 4.4 (CE) is being planned/developed:

Puppy 4.4 CE - Phase 1: pet tests
http://murga-linux.com/puppy/viewtopic.php?t=46941

What shall I work on first for the next CE release?
http://murga-linux.com/puppy/viewtopic.php?t=47065

Puppy 4.4CE
http://murga-linux.com/puppy/viewtopic.php?t=46497

T2 8.0rc binary packages

I have uploaded the packages that I compiled in the T2 build system, to here:

http://bkhome.org/binaries/t2/

508 packages, occupying about 445MB.

The next release of Woof will be able to build a distro from these.

Satellite TV: NITV

I reported recently that I was thinking of installing satellite television:

http://bkhome.org/blog/?viewDetailed=01118

Given that I can only get one terrestrial TV channel, our ABC (Australian Broadcasting Corp., Government non-commercial), and that very weak, I reckoned it was time to go for free-to-air satellite TV.

I know in some parts of the world, satellite TV is the normal thing. For me it's a big deal. Most Aussies are well-served by the terrestrial channels.

So, on my last trip to Perth I bought the parts. There's this neat little hand-held "satellite finder", just a needle gauge and beeps -- the gauge shows signal strength and the pitch of the beeps goes up, and the beeps go faster, as the signal increases. It was a cinch to get the dish pointed at the Optus C1 satellite. It only cost AU$15:

http://www.sciteq.com.au/store/product_info.php?products_id=149&SciTeqsid=139c3f4d60243d27b11e5b70686703ba

After getting it all setup, I then had to go onto the Internet to activate the Optus smartcard -- oh, "two business days". Today is Saturday. Oh well.

The Optus C1 satellite only has two un-encrypted channels, Optus Sydney which only has a test pattern and a continuous tone, and NITV. You can also see NITV online:

http://nitv.org.au/

...I have been sitting here watching it. Hmmm, after the initial novelty perhaps even the most keen indigenous person would be changing to some other channel that provides entertainment. ...I don't mean that comment to come across as disrespectful.

This is the receiver I bought:

http://www.sciteq.com.au/store/product_info.php?products_id=388&SciTeqsid=139c3f4d60243d27b11e5b70686703ba

Not knowing anything about these things, I just bought what the guy behind the counter recommended. One nice feature, a USB socket that will take a drive, so can save to it.

Quirky page

People are going to be asking, what is Quirky? I know Lobster has created a wiki page, but I thought that I should put something on my own Quirky page:

http://bkhome.org/quirky/

Sqlite 3.6.20

A note for developers...

My T2 compile that I'm doing for Quirky (that's still chugging along) has Sqlite 3.6.20. However, compile failed as it complained that it needed 'tclsh', even though I had specified '--disable-tcl'.

I rolled back to Sqlite 3.6.12, that was in a T2 compile I did in April 2009. That does compile without Tcl.

I joined the Sqlite mail-list and asked about this. I was advised to use the 'sqlite-amalgamation-3.6.20.tar.gz' source. Sure enough, that compiles without Tcl.

Get it from here:
http://www.sqlite.org/download.html

The "amalgamation" is explained here:
http://www.sqlite.org/amalgamation.html

I have compiled SeaMonkey 2.0 in T2. Note that it can use the system Sqlite, but it must be version >= 3.6.16. Earlier SeaMonkeys have an internal Sqlite, so this saves space.

N

Dim version control

I think that I have finally found a version control system that is to my liking:

http://www.oligem.com/dim/

With all due respect to CVS, SVN, GIT and the rest, I find them too complicated and big.

Dim is a single shell script, and even works with Dash/Ash.

I'm thinking that I'll have Dim in the 'devx' SFS file, so it becomes a standard feature in Woof builds.

I reckon, finally, I might put Woof onto version control on my website, and let others have direct input.
...if anyone sees a problem with doing that using Dim, let me know.

T2 compile in progress

I decided to do a T2 compile from source, as a basis for Quirky Linux. Mainly because I want it to be small.

My PC that I have dedicated to T2 builds is chugging away...

The main T2 developer, Rene, recently announced version 8.0rc, so they are finally moving toward another release. I grabbed the build tarball from subversion a few days ago.

The SVN package in Puppy's devx is not up to it, even though it was compiled in T2. It lacks support for connection to https://, which is due to a missing dependency. That's T2's fault, not including that dependency -- which is quite odd.
I used my Upup to grab T2 from SVN.

However, the SVN download still didn't work, something was missing. I solved the problem by merging my SVN download into 't2-8.0rc.tar.gz' (the official 8.0rc tarball).

It's not for the faint-hearted. Lots of compile errors, but I'm getting there.

T2 home: http://t2-project.org/

Decisions, decisions...

I can't make up my mind whether "barryk.org" or "bkhome.org" is better.

The latter might convey a slightly more professional impression maybe?

I have registered both of them for two years, but should settle on just one for preferred use.

Raffy, I know there are other possibilities. Strangely enough, I actually want a domain name for which all other variants are already registered. That is, I have .org, and all the other TLDs are already registered (bkhome.com, etc). My reasoning is, it is good to know what my neighbours are before I "buy in".

Notes for 4.4 developers

Just a couple of quick notes, while I think of it.

OpenSSH package in 4.3.1 is too old (4.7p1). It doesn't support SSH v2. My host Hostgator.com recently upgraded their SSH, and I had to compile the latest OpenSSH (5.3p1).

Pburn. I'm using version 3.1.2, in Puppy 4.3.1. I burnt an audio file to CD. At completion I exited via the "File" menu, that is, the proper way. However there was still a process left running. I had to kill it to be able to unmount the partition. This problem is repeatable.

Test post, barryk.org

This blog has a new URL, barryk.org/blog. First test post.

Actually, I copied the blog, and the old one at puppylinux.com/blog is still running. Next step is to figure out how to do some URL-rewriting to redirect to the new blog.

Quirky idea #1 works

I had already done a small test, now I have written a script in Woof, '4quirkybuild', to do a full build of Quirky.

I think that I discussed this idea with Raffy a long time ago. Well, it was briefly mentioned.

Puppy has at least three files, vmlinuz, initrd.gz and pup-xxx.sfs.

The initramfs, or to use it's old name "initial ramdisk", is initrd.gz. After the kernel vmlinuz has loaded, it then loads this initramfs, which is a mini Linux working environment. It does the searching of the drives to find the Puppy files, and after having found pup-xxx.sfs (and maybe the 'pupsave', 'zdrv', and extra sfs files) then sets up the layered filesystem and peforms a 'switch_root' into this layered filesystem.

Puppy can be built with the pup-xxx.sfs inside the initrd.gz, which is good because then Puppy doesn't have to look anywhere to find it. It does slow bootup though, as it all has to be loaded into RAM.
It is also simpler to setup booting, especially over a network, as there are only two files.

Now, the quirky idea takes this one step further. It is possible to compile the kernel vmlinuz with the initrd inside it. So, if we build the pup-xxx.sfs inside initrd, then we compile the kernel with the initrd inside the kernel... we end up with just one file!

Furthermore, the 2.6.30 kernel inroduced lzma compression. This reduces the size of the kernel, but we can keep plain gzip compression for the pup-xxx.sfs file (which runs faster). Though, I don't think that applying lzma compression to a sfs file reduces it much.

Anyway, I did a test build, based on 4.3.1 pet packages, and created 'quirky000'. This is a Linux kernel, with, as I explained above, the initrd and the pup-xxx.sfs inside it.
I put it on a USB Flash drive, and modified the 'syslinux.cfg' file to just this:

default quirky000 pmedia=usbflash

...the pmedia parameter isn't really needed.

Yep, it works! booting is a bit slow, as everything loads into RAM. But, this is the direction that I am aiming Quirky -- a distro that runs in RAM only.

Hey, this is fun!

This idea #1 is going to be very useful for those who want to do network booting, I reckon.

Quirky is coming

I have had some ideas in the back of my mind for sometime, but didn't know if they would work. These ideas are, well, "quirky", hence when I create a distro based on them, then I'll name it "Quirky".

I conducted tests on two of these, and they work. Three of the ideas tackle these areas:

1. Simpler installation and booting.
2. More security when running.
3. Clean shutdown.

I needed to make configuration changes to the kernel, so took the opportunity to grab the latest, 2.6.31.5.

This evening I'll be figuring out how to build Quirky using Woof. I'll probably have to write an extra script, 'quirky_build' or something.

Why do I bother?

I sometimes wonder why I bother. In my position, I deal with criticism on a daily basis. Some of it comes direct to me in emails and pm's.
Then there is negativity by some who lurk on the #puppylinux IRC channel.
Then there are those who post to the forum with superficial, mis-informed, nit-picking, and just plain wrong comments, like in this thread:

http://murga-linux.com/puppy/viewtopic.php?t=48160

I really don't need the aggravation. The latest contribution on the forum has prompted me to think some more about my retirement.

My original plan was to retire from Puppy development early in 2010. I want to keep working on Woof and do some Genie coding, and that will likely continue into 2010, 2011.

The multi-user improvement -- yes that is a definite contender for incorporation into Woof. That's a short-term goal, I might do that soon.

Puppy 4.4 and 5.0 are now being developed by the community. That's good. I'll probably give the access to the ibiblio.org site to one or two community members soon.

I won't be bringing out another official Puppy release, so I am now retired in that regard.

What I might do is release a new distro sometime, which is something that I have mentioned before. It will be 100% personal, no irc, no forum, just my thing. In fact, I'm already working on it.
It is not intended to be a general purpose distro.

Multiuser Puppy

This needs to get lots of attention I think, as pizzasgood has put in considerable effort to solve the problem of running Puppy with multiple users:

http://www.murga-linux.com/puppy/viewtopic.php?t=47409
http://www.murga-linux.com/puppy/viewtopic.php?t=47410
http://www.browserloadofcoolness.com/

What I commend in particular, is the effort that pizzasgood put in to document everything.

It's good stuff, and I'm looking into incorporating this into my Woof build system.

Genie bugs reported

I'm working through my Genie web pages, updating them, and of course testing all the examples.

I am using vala from git, October 24, 2009.

Two of the examples (so far) do not compile and I have reported to the bugzilla:

https://bugzilla.gnome.org/show_bug.cgi?id=599635
https://bugzilla.gnome.org/show_bug.cgi?id=599636

Main Genie page updated

Now that I am embarking on furthering my knowledge of Genie, I have made a start by updating my main Genie page:

http://puppylinux.com/genie/

Problem with GTK 2.18.3

I have got a problem in Karmic Puppy. If I right-click on a file, choose something from the menu, the file-thumbnail becomes highlighted and stays highlighted. I have to click elsewhere in the Rox window to deselect it.

This is a feature of GTK 2.18.3, the version used in Karmic Koala. I confirmed this by compiling GTK 2.18.3 in Puppy 4.3.1, which I'm running right now.

This is very unexpected behaviour, and extremely annoying. I thought it best to post this, as others will encounter it, and you need to know that it is a GTK problem. The thing is, is this behaviour configurable? How?

By the way, I compiled glib 2.22.2, pango 1.26.0 and gtk+ 2.18.3 in Puppy 4.3.1, and I'll make them into PET packages and upload them to ibiblio. I'll also make database entries for them in Woof.

So, 4.4 developers can use these packages if they want, or they may already have compiled these latest packages.

Karmic Puppy live-CD

Get it from here:

http://puppylinux.com/test/karmic-puppy-20091024/

The Woof tarball and the "Karmic Puppy" live-CD ISO file are just snapshots of work-in-progress.

The Woof tarball is currently configured to build from Ubuntu Karmic Koala packages, and I used 9.10RC packages. The Intel graphics on my laptop now seems to be behaving itself, no flicker!

I think that it should be pretty straight forward to get Woof to build a Dpup as well, and ttuuxxx and gposil may have use for this latest Woof in their Dpup project. Anyone is welcome to use Woof and Karmic Puppy and further develop them.

Their Dpup project:
http://murga-linux.com/puppy/viewtopic.php?t=47592
http://www.dpup.org/index.html

But for me, well, I'm taking a break, going to work on learning more about Genie and update the online docs. My main Genie page was last updated in December 2008 -- well overdue for further attention!

ROX-Filer 2.9.0

I have compiled the latest version in Puppy 4.3. I want to use it in Karmic Puppy, compiling it in 4.3 ensures that the same package will work in any build from 4.3 upward.

I did compile this awhile ago, but had problems with the mime types -- the thumbnails did not display properly. Ttuuxxx solved that, and I looked at ttuuxxx's Rox package -- yes there are name changes in some mime files. I now have it working properly. Thanks ttuuxxx.

Shared-mime-info package is now upgraded to the latest, 0.70. The 'dev' component now goes into the 'devx' SFS file.

Woof: Making 0setup faster

I reported recently on the super-fast code that Wosh has created:

http://puppylinux.com/blog/?viewDetailed=01167

Wosh sent me version 0.0.1 of PPKG today, and it is nice but not yet quite what I need for building Karmic Puppy.

To fill the immediate need, I thought that I would take another look at the Debian/Ubuntu-to-Puppy database conversion code in the '0setup' script.

Testing the Ubuntu "main" repository db, the script took about 11 minutes. I applied some lateral thinking, and have now got it down to about 2 minutes.

Here is the new code:

v1PATTERN="s%\\-[0-9.]*${DISTRO_BINARY_COMPAT}[0-9.]*$%%"

v2PATTERN="s%\\+[0-9.]*${DISTRO_BINARY_COMPAT}[0-9.]*$%%"

   while read DB_ONELINE

   do

    eval $DB_ONELINE

    case $DB_ONELINE in

     Description*)

      DB_fullfilename="`echo -n "$Filename" | rev | cut -f 1 -d '/' | rev`"

      DB_path="`echo -n "$Filename" | rev | cut -f 2-9 -d '/' | rev`"

      DB_version="`echo -n "$Version" | sed -e 's%^[0-9]:%%' -e "$v1PATTERN" -e "$v2PATTERN"`"

      DB_pkgrelease="`echo -n "$Version" | sed -e 's%^[0-9]:%%' -e "s%${DB_version}%%" -e 's%^\\-%%' -e 's%^\\+%%'`"

      DB_pkgname="${Package}_${DB_version}"

      DB_category="`${FIND_CAT} $Package "$Description"`"

      DB_dependencies="`echo -n "$Depends" | sed -e 's%, %\n%g' | cut -f 1 -d ' ' | tr '\n' ' ' | sed -e 's% $%%' -e 's% %,+%g' -e 's%,$%%'`"

      [ "$DB_dependencies" != "" ] && DB_dependencies='+'"$DB_dependencies"

      echo "$DB_pkgname|$Package|$DB_version|$DB_pkgrelease|$DB_category|${InstalledSize}K|$DB_path|$DB_fullfilename|$DB_dependencies|$Description|" >> $ONE_PKGLISTS_COMPAT

      echo -n "$Package "

     ;;

    esac

   done<<_END1

$(grep -E '^Package:|^Installed\-Size:|^Architecture:|^Version:|^Depends:|^Filename:|^Description:' ${ONE_PKGLISTS_COMPAT}pre | tr '[\t"|]' ' ' | tr -s ' ' | sed -e 's%^Installed\-Size%InstalledSize%' -e 's%: %="%' -e 's% $%%' -e 's%$%"%')

_END1

Of course the big one is the "universe" repo. The database file to be processed is 24MB, a lot to churn through. Universe now takes 9 minutes, so all-up the conversion time will be about 12 minutes, apart from the time to download the Debian/Ubuntu db files.
The Debian repo is a bit smaller so conversion will be quicker.

This faster script is just a temporary fix, as I expect to move onto PPKG soon.

Aghanistan: why are we there?

The Afghanistan election fraud is in the news right now, but I watched a TV documentary a couple of nights ago, and the situation is very depressing...

"Four Corners" is a long-running and very reputable documentary program here in Australia. The last one was on what has happened to all the aid going into Afghanistan:

http://www.abc.net.au/4corners/content/2009/s2714822.htm

18 billion dollars. The reporter wanted to follow the money trails, find out where it went.

The UN claims that 268 schools have been built, and the reporter asked for some addresses. But, he had trouble getting even one address, and when they finally gave him one, he went there and found a billboard with a picture of a grand building -- the billboard had been put up two years ago. What was actually built? -- there were some tents and the students were studying in those.

And that was a lucky school. he found another "school", they didn't even have a tent, they just sat outside, until it snowed too much.

Then there was the hospital. The money donated was, if I recall, about 2.5 million, of which foreign NGOs took most for admin (etc.) and under 1 million finally got to the local construction company. The final building is shoddy, falling down, yet the foreign quality control engineers gave it the thumbs up.

So, there is virtually no reconstruction, and whenever the reporter spoke to local people there was heaps of bitterness.

Ah, but there is one area where there's lots of activity -- mansions for the rulers. No delays there. Also no explanations how they can afford million dollar mansions.

In summary, everyone is on the take, both Afghanistan and foreign companies. The UN, well, I don't know what to say about them.

The overall picture is that the hearts and minds of the people have not been won. Just the opposite. If that's the case, then the foreign troops are staying there with some other agenda.

Howto compile apps

It is happening slowly, but I have updated another web page:

http://puppylinux.com/development/compileapps.htm

Another ibiblio mirror

Tony sent me this message:

Hi Barry,

just to inform you I maintain a puppy mirror here since couple of years at ...

Goodie Domain Service - www.gdsw.at (frontend)
Vienna University of Technology, Austria, Central EU
located @
ftp://gd.tuwien.ac.at/linux/puppylinux/
http://gd.tuwien.ac.at/linux/puppylinux/

mirrored from ibiblio

I'll add this to the download page.

Karmic Puppy: first build

I'm running "Karmic Puppy" right now. This is Puppy built from Ubuntu Karmic Koala packages.

Working nicely!

One thing though, most annoying, have got that periodic screen flicker. This happens with Intel video hardware and arrived with Xorg 7.4.
I did manage to get rid of it in a Jaunty build, by using a later Intel Xorg driver, but later when I tested an early version of Karmic, the flicker was back. Now, with the final release of karmic due out on the 29th of this month, the flicker is still there.

Anyway, apart from that, this pup is running fine. A bit fatter than your usual pup of course, with all those dependencies -- this build is 113MB with the SCSI and big-analog-modem drivers removed.

There are a few things that I want to improve, which I'll work on for a few more days.

I might look into compiling a Intel Xorg driver out of subversion.

Woof plans

Just to let everyone know what I'm doing, or rather, plan to do.

For sometime I have only used Woof to build a "Ppup" (Puppy from Puppy packages), that is, working toward 4.3.1.

So, the other compatible-distros have not been recently tested, and with the GUI not at all.

I have now started working on building from Karmic Koala (the upcoming Ubuntu 9.10) packages. Immediately I have found some problems with the GUI. Also I have made some improvements to the GUI to clarify how to use it.

When I get a successful "Karmic Puppy" built, I'll upload another snapshot of the Woof tarball.

Puppy 4.3.1 released

Took a long time to upload to ibiblio.org, but finally got there! Download from ibiblio only for now, as not yet filtered through to the mirrors:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/puppy-4.3.1/

When the mirrors catch up, you can get to them via the main Puppy download page:

http://puppylinux.com/download/

There are lots of ISOs to choose from:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/puppy-4.3.1/readme-files.htm

...the main guy, that will suit most people, is 'pup-431.iso', size 104.9MB.

The main announcement and release notes page is here:

http://puppylinux.com/download/release-4.3.htm

...scroll down that page and you will see the list of differences between 4.3 and 4.3.1.

I have started a forum thread for bug reports and feedback on 4.3.1:

http://murga-linux.com/puppy/viewtopic.php?t=47563

...I will monitor it, but my attention will now be moving to development of Woof. Forum member technosaurus is carrying on with developing the Puppy 4.x series, that is, 4.4, and you can find various threads about it on the forum.

Preparing for 431-final

I decided to go back to my Partview. I like what Ecube has done, but I personally want to be able to see the free space in all partitions, mounted or not. Ecube's script could be extended to do that. My script is rather slow, as it generates gif images for each partition.
I have applied clarf's bugfix so that my partview works properly in older kernels without libata drivers.

Forum member MinHundHettePerro reported the problem of icon-stacking when there are too many drive icons to fit horizontally on the screen. This will have to be tackled in a future Puppy. The appropriate script is /sbin/pup_event_frontend_d, function free_coord().

Zigbert has upgraded Pburn to 3.1.2-1, incorporating a fix from rerwin.

Zigbert has upgraded Pfind to 4.15-1, incorporating a fix from rerwin.

One thing that is disappointing is that there have been a lot of bug reports using the Network Wizard, but we currently don't have anyone maintaining it. This will need to be tackled for 4.4.

If I recall rightly, there have been two main problems: having to scan twice for a wireless network, and settings not (properly) remembered between boots.

To a lesser extent problems have been reported with Pnethood, and again it is not currently maintained.

Puppy 4.3.1-RC2

Earlier today I was unable to login to ibiblio.org, so I uploaded the files to here:

http://puppylinux.com/test/puppy-4.3.1-rc2/

Ah, uploads at 3 times the speed of ibiblio (30KB/sec compared with 10KB/sec)!

Just a minimum upload, only one ISO, 2.6.30.5 kernel, no Intel 536/537 modules, 104MB.

I have also uploaded Woof, so anyone who is interested can try the new GUI.

Right now it is 6pm Wednesday October 14 here in Western Australia. I'll be busy with other stuff all day tomorrow, so about 24 hours from now I'll take a look at any feedback.

Note, I put in Ecube's Partview, although it does have a bug, doesn't run if no partitions are mounted. It only displays free space in mounted partitions, whereas my Partview in 4.3.1-RC1 displays free space in all partitions. Actually, it would be possible to extend Ecube's Partview to also display free space in unmounted partitions. My script is still there, renamed to 'partview_all'.

My previous blog post summarises the bugfixes:

http://puppylinux.com/blog/?viewDetailed=01168

Forum thread for feedback:

http://murga-linux.com/puppy/viewtopic.php?t=47795

4.3.1 RC1, RC2 bugfixes

Here is a summary of the bugfixes so far, since the release of 4.3:

4.3.1-RC1
New modem drivers and improved modem detection and dialup, from rerwin.
Fixes for CD Remaster script.
Asunder CD ripper replaces Ripoff (which crashed).
Cdparanoia upgraded to latest (previous crashed).
You2pup, fix for spaces in paths.
Ayttm multi-protocol chat client, upgraded to 0.6.0-9.
DidiWiki personal wiki upgraded to 0.8.
JWM window manager upgraded to revision 457.
NicoEdit, our secondary text editor, upgraded to 2.4.
Pburn upgraded to 3.1.1.
'resolv.conf' circular symlinks maybe fixed?
JWM Configure tool bug fix ('.jwmrc-tray' got corrupted).
'man' and Help page fixed when search on linux.die.net.
Frequency scaling fix for "small" iso (modules were missing).
Shutdown problem when upgrade 'pupsave' (shutdown scripts in wrong place).
Partview, a new tool to view free space in drives.
Sylpheed PET package upgraded to 2.7.1 (previous crashes).
Mouse Wizard fixed (showed USB when a PS/2 was connected).
Lots of bugfixes for the Puppy Package Manager (PPM).
Some corrections in the locale-choosing dialog.
Boot from ext4 and save 'pupsave' to ext4 partition fixed.
The famous "insert key" crash in Mozilla browsers fixed.
Fixed modules loading for kernels prior to 2.6.24.

4.3.1-RC2,-final
Some SVG images were not displaying in GTK applications. Fixed.
Typo in You2pup script.
CD/DVD Drive Wizard was broken. Rewritten.
Fix for JWM Config keyboard script.
Grammar checker default in Abiword is now off, as it is so slow.
Fix for overlapping drive icons on desktop at certain screen resolutions.
Partview simplified, faster.
Burniso2cd could sometimes fail to perform the verify-step properly. Fixed.
Pburn bugfixes.
snd-ali5451.ko module added to Smartlink modem driver system.

I'm uploading 4.3.1-RC2 now, will announce it soon.

Debian to Puppy database converter

Brilliant! Look at what Wosh has done:

http://www.murga-linux.com/puppy/viewtopic.php?t=47763

I don't know about the need for a contest though, I reckon that 10,553 times faster than my script is pretty good

I want to get back onto testing Woof with other distros, probably Lenny and Karmic in the next few days, well immediately after I get 4.3.1-final released.

So your converter is so so welcome. That script '0setup' is not just used in Woof, the 3builddistro script copies 0setup to /usr/local/petget in the built Puppy. It is used by the Puppy Package Manager to update the package database.

So, when a user goes into the 'Configure PPM' section and clicks on the button to update the package databases, they don't want to be waiting around for one hour! That's the main reason your program is so important.

I'll get onto playing with it real soon.

Note, I think that I filtered out []() from the description fields as those characters may have interfered with grep or sed when reading and processing the database. But, would have to check that, probably there is a workaround for that. Or, I may have just filtered them out because I thought potentially they could cause trouble.

One other small detail, "menu categories" are really just "categories". A package can belong in a certain category, say "System" but not have a menu entry. The categories are mostly for the Puppy Package Manager to group the packages.

SVG solution

Thanks to forum member ljfr, I was able to implement a solution to the problem of some SVG icons not displaying (see previous blog post).

ljfr found that upgrading the shared_mime_info package to the latest fixed the problem.

In Puppy, the shared_mime_info files are cutdown and inside the rox_filer package. It is also a rather old version of shared_mime_info.

shared_mime_info is overdue to be upgraded, and will have to be done when we upgrade rox_filer.

But for now, with 431-final imminent, I do not want to "rock the boat". I found out that just one file needs to be upgraded, /usr/share/mime/magic, so as a temporary measure, just for 431 really, I have created a new package 'zz400_mime_hack-431.pet' with just that one file in it.

Note, the reason it starts with "z" is so that Woof will copy it into the live-CD last, overwriting the existing 'magic' file.

I have done some tests, and nothing else seems to be broken. The Rox mime-handling uses the 'magic' file, as does the 'file' utility, and they seem happy with the new file.

But, I think that I will have to bring out an RC2, just to be sure. Of course, once again, I hope that I can just promote it to the final after a couple of days.

I will probably upload 431-RC2 tomorrow.

SVG mystery

4.3x testers have reported that some SVG images do not display in ROX or Viewnior.

Those in /usr/local/lib/X11/themes/FlatOrange do.

Newer ones, such as created in Inkscape, do not.

I wondered if perhaps 'librsvg', which is the package that provides SVG support for GTK, needed to be compiled with its optional dependency 'libcroco' (provides support for SVG images that have CSS formating).

So, I compiled libcroco 0.6.2 and recompiled librsvg 2.22.3. Nup, same problem.

So, I grabbed the latest librsvg, 2.26.0, but still no good.

Then I tried this ('rsvg-view' is a utility in the librsvg pkg):

# rsvg-view

...hey, it displays!

So, librsvg itself is okay, GTK is somehow the problem.
If I try to open one of the newer SVG files in Viewnior, it reports an "unrecognised file format". C**p.

So, I wondered, just what is different between the SVG files that do display, and those that don't?

I have one that doesn't, 'brasero.svg', the first few lines of the file:





   xmlns:svg="http://www.w3.org/2000/svg"

   xmlns="http://www.w3.org/2000/svg"

   xmlns:xlink="http://www.w3.org/1999/xlink"

   version="1.0"

   width="48"

   height="48">

       id="defs3">

           id="linearGradient5349">

/usr/local/lib/X11/themes/FlatOrange/camera.svg works, here are the first few lines:



"http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd">



   version="1.0"

   x="0.0000000"

   y="0.0000000"

   width="162.50000"

   height="162.50000"

   id="svg183"

   sodipodi:version="0.32"

   inkscape:version="0.36"

   sodipodi:docname="/usr/local/lib/X11/themes/Outline-svg/camera.svg"

   sodipodi:docbase="/usr/local/lib/X11/themes/Outline-svg"

   xmlns="http://www.w3.org/2000/svg"

   xmlns:inkscape="http://www.inkscape.org/namespaces/inkscape"

   xmlns:sodipodi="http://sodipodi.sourceforge.net/DTD/sodipodi-0.dtd"

   xmlns:xlink="http://www.w3.org/1999/xlink">

So guess what? I copied that "DOCTYPE" declaration from camera.svg, pasted it into brasero.svg -- now brasero.svg displays!

My memory of the SVG standard is that it needs that DOCTYPE declaration. So how come all the recent SVG images I have downloaded, created by Inkscape, don't have it? I just checked something, compiled GTK 2.16.6, still the same problem.

So, I am stuck with a mystery. I have a workaround, but really when we download SVG images we want them to display as-is, not have to edit them. Anyone got a solution? As things stand, I will have to go to 431-final with this not properly resolved.

Hard-Puppy page updated

I have updated this page:

http://puppylinux.com/hard-puppy.htm

Web pages updated

There is a lot more to do, but I have made a start updating web pages at puppylinux.com.

A page that is highly recommended that all Puppy newcomers should read is the overview of package management:

http://puppylinux.com/development/package-management.htm

I have also updated, at least partially, pages on Woof and the PPM:

http://puppylinux.com/woof/
http://puppylinux.com/woof/ppm.htm

This page is now a redirect to the main Woof page:
http://puppylinux.com/development/puppy-unleashed.htm

Documentation on the old Unleashed and PETget have been replaced with Woof and the PPM.

Service Pack for 4.3

Get it from here:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/puppy-4.3.1-RELEASE-CANDIDATE/service_pack-430-to-431.pet

This is not as "complete" as a proper upgrade to Puppy 4.3.1, but is handy in certain circumstances, such as you have a full install of 4.3 to hard drive -- in that case, just download and install the PET and you have upgraded.

At least I hope so, I haven't tested it.

Probably the Service Pack is adequate if you are running 4.3 from CD, USB or frugal install and you just want a quick upgrade to 4.3.1.

One thing that is missing is the latest 'initrd.gz', needed for booting from ext4 partitions. But you don't need that for a full hd installation. But you will have to get it for your USB/frugal installation -- but then you might as well grab the new pup-431.sfs too.

But of course the Service Pack is small, just 4.4MB, whereas pup-431.sfs is about 103MB

Note that I did not put the 537* modem drivers into the service pack.

Puppy 4.3.1-release-candidate

Available from here:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/puppy-4.3.1-RELEASE-CANDIDATE/

Puppy 4.3.1 is a bug-fix release of 4.3. Here are the highlights of things fixed, upgraded and improved. These items are in my blog (read the archive for October http://puppylinux.com/blog/?do=archive):

New modem drivers and improved modem detection and dialup, from rerwin.
Fixes for CD Remaster script.
Asunder CD ripper replaces Ripoff (which crashed).
Cdparanoia upgraded to latest (previous crashed).
You2pup, fix for spaces in paths.
Ayttm multi-protocol chat client, upgraded to 0.6.0-9.
DidiWiki personal wiki upgraded to 0.8.
JWM window manager upgraded to revision 457.
NicoEdit, our secondary text editor, upgraded to 2.4.
Pburn upgraded to 3.1.1.
'resolv.conf' circular symlinks maybe fixed?
JWM Configure tool bug fix ('.jwmrc-tray' got corrupted).
'man' and Help page fixed when search on linux.die.net.
Frequency scaling fix for "small" iso (modules were missing).
Shutdown problem when upgrade 'pupsave' (shutdown scripts in wrong place).
Partview, a new tool to view free space in drives.
Sylpheed PET package upgraded to 2.7.1 (previous crashes).
Mouse Wizard fixed (showed USB when a PS/2 was connected).
Lots of bugfixes for the Puppy Package Manager (PPM).
Some corrections in the locale-choosing dialog.
Boot from ext4 and save 'pupsave' to ext4 partition fixed.
The famous "insert key" crash in Mozilla browsers fixed.
Fixed modules loading for kernels prior to 2.6.24.

Test like crazy! I hope to upload the final in a few days. It may even be the case that the release-candidate is fine and can simply be promoted to final.

I will upload a Service Pack for Puppy 4.3, for those who have that version installed and would just like to install the fixes without doing a full version upgrade.

I am doing a little bit more work on the Woof GUI and it will be in Woof when 4.3.1-final is released. The GUI is not complete and will have many more features, but should be basically usable. So far only tested building Puppy 4.3.1.

I created "delta" files, so if you have previously downloaded 'pup-430.iso' and 'devx_430.sfs' then there are small .delta files that can be used to upgrade -- see "Xdelta file difference manager" in "Utility" menu.
The delta file for the ISO is a bit big (38MB), due I think, to a different organisation -- I built 431 without a 'zdrv' SFS file.

Forum thread to report bugs:

http://murga-linux.com/puppy/viewtopic.php?t=47671

The 'PUP_430.SFS' problem
Many people who used Isobuster in Windows had this problem. It is solved by upgrading to the latest version of Isobuster (which is able to read Rock Ridge extensions in the ISO file).

Also, to make it easier for people who want to open up the ISO, I have uploaded the individual files to 'inside-iso' folder on ibiblio.

Intel 536/537 analog modems
I left these out of 'pup-431.iso' due to their huge size. Instead, I did a special build, pup-431-INTEL_MODEMS.iso' in folder 'special-puppies'.

431 release candidate soon

I'm uploading it now. I was putting it together about 1am, but so tired I was making mistakes... I think I built the iso ok, started the upload then went to bed. Now it's 6am, still uploading.

I'll let you know when it's done. Still another 4 - 5 hours.

I'll also prepare the Service Pack for 430 today.

Final stuff before 4.3.1

A quick note:

I found the bug that caused the Remaster CD script to crash if there are two optical drives.

I have applied rerwin's latest ltmodem stuff, pupscan change, pmodemdiag PET pkg into puppy4 repo.

4.x plans

I plan to upload 4.3.1-release-candidate in a couple of days.

There are still some issues in 4.3 that are not resolved, so don't expect miracles, but also there are a lot of bugs fixed.

The idea of the release-candidate is to check that none of my "fixes" broke something else. If there's nothing dumb in it, or something is found that I can readily fix, then the final should hopefully follow a few days later.

Then I'm going back to Woof and working on the GUI plus various usability enhancements.

Technosaurus will be full steam ahead on 4.4... which leads me to technosaurus' initiative, to put Puppy development onto google:

http://murga-linux.com/puppy/viewtopic.php?t=47627

...check it out, give feedback to technosaurus.

Asunder 1.9

There have been reports that Ripoff, the CD song ripper in 4.3 and earlier puppies, dies half-way through if you try to rip an entire CD.

Looking at the Ripoff project page, I see that it has not been updated for a long time.

On the otherhand, Asunder is very active, so I compiled the latest, version 1.9, and tested it. Yep, lovely, I ripped off an entire CD to MP3 files.

So, 4.3.1 will have Asunder, not Ripoff.

Asunder home:
http://littlesvr.ca/asunder/

cdparanoia III 10.2

It has been reported that cdparanoia crashes in pup 4.3.

That is a rather old package, III-alpha9.8, which dates back to 2007. I have compiled the latest, III-10.2.

Aqualung improvement

Here is something else that I have found in ttuuxxx's "431.1 Firefox Puppy" thread....

Forum member Gedrean posted this:

Only problem I had was, and not sure if this is with yours or with puppy 430 in general, that Aqualung would never actually open with a file I clicked on, I had to afterwards drag it in there.
Also the Aqualung script, when I read it, seemed horribly inefficient. Kill the other processes, then relaunch, with the argument? Not too elegant either.

So I had about 5 hours of editing, finding that behaviors would change from minute to minute (probably my corrupted computer somehow), and agonizing over the Aqualung command line options and their strange intricacies (sometimes they will behave as advertised, ONLY if other arguments exist) but here's what I've done:

I've managed to make the default behavior for Aqualung to Enqueue (add to playlist without clearing rest of it), caused it to only ever open ONE process and not need to kill others, and set it up so that there's a right-click for directories to enqueue the whole directory and contents in as well! (Great for my huge music directory!)

I've got three files I'll paste in code, they're very short, and I'll detail what I did (something I wish others would do, as bash scripts are incredibly hard to read!)

First, aqualung launcher script, located in /usr/bin :
Code:
#!/bin/sh
exec aqualung2 --session=0 -E /dev/null "$@"

--session=0 tells it to do remote cue functions (according to the aqualung usage text) with session #0, which is either the first one already opened (and in this case the only) or the one which is opening now if no other session exists.
If you'll note, especially ingenious, it attempts to Enqueue /dev/null, which will fail out with an error that it is unable to open that file ... this is required so that the -session=0 will actually do anything other than opening a new session.
-E tells it to enqueue, and then it adds the remainder of any files. I spent a good amount of time making it add files with unicode/utf-8 chars, and spaces, and goofy characters, and it all works!

Next, defaultaudioplayer script, located /usr/local/bin, and as an added bonus this is the same code for the Add to Audiolung Queue script located in ~/.config/rox.sourceforge.net/OpenWith/.inode_directory :
Code:
#!/bin/sh
exec aqualung "$@"

Of course, aqualung2 is the name of the aqualung app itself, sitting in /usr/bin.

Thought I might share these and see what people (you, others?) think, and maybe contribute a bit more than requests to the whole puppy thing.

I have done some basic testing, and this looks like a very nice improvement, a big step toward making Aqualung usable. It prevents multiple instances of Aqualung starting everytime a sound file is clicked on.

I have implemented Gedrean's excellent fix, except named the startup script 'aqualung_wrapper' and left the executable as 'aqualung'.

Note:
One thing in Aualung that seems so unintuitive to me: if you have a playlist, click on one file in the list to highlight it, then click the 'Play' button, wouldn't it be nice if just that file played? (and not the whole lot)

UPDATE:
I found a a later post from Gedrean, a different script to fix a problem when defaultaudioplayer is called from streamtuner.
However, I have found showstopper problems with both scripts. The second one, files no longer played (aqualung did not even start) when I clicked on a sound file in Rox.

So, I have rolled back to the previous aqualung package, not using Gedrean's fix.

You2pup fix

dio444 fixed a bug in trio's You2pup:

There is a bug that causes the script to fail, and hence the entire You2Pup system to fail if a local filename with spaces in it is chosen. The fix is easy, simply add "" to a few arguments.

I found this fix in ttuuxxx's "431.1" forum thread (430 with fixes and various changes including Firefox).

Hmmm, good thing I decided to scan through that thread.

Ayttm 0.6.0-8

Technosaurus has compiled the latest from revision control. Thanks also to Michalis:

http://murga-linux.com/puppy/viewtopic.php?t=42959

I have upgraded to technosaurus' package.

Note to technosaurus: You need to include the *.la files in /usr/lib/ayttm. Ayttm needs them to recognise the plugins. At least, it used to.
And of course, if you were making an official package to replace mine, you would have to include the 'pinstall.sh' script.

DidiWiki 0.8

It's great, forum member 'amigo' has done a lot of work on DidiWiki, our personal Wiki. This was a dead project. Forum thread:

http://murga-linux.com/puppy/viewtopic.php?t=47185

I have upgraded to the latest, version 0.8.

JWM revision 457

Pstef posted to this blog yesterday:

Joe has updated JWM on its SourceForge site about two weeks ago. Those updates include Patriot's contribution too.

I'm keeping changes to a minimum right now, as 431 is imminent, but this is too much of a temptation. So, I have sneaked this package upgrade in too.

NicoEdit 2.4

I have upgrades Nicolas's NicoEdit text editor:

http://www.murga-linux.com/puppy/viewtopic.php?t=46629&start=15

Pburn 3.1.1

I have upgraded zigbert's Pburn to the latest.

http://www.murga-linux.com/puppy/viewtopic.php?t=23881

resolv.conf circular symlinks

Forum memeber disciple posted this to the forum recently:

This is a problem that seems to have been around for at least four years, although I think normally the people who report it are connecting with adsl modems rather than through an ethernet card to a router with DHCP.

The last time I saw it reported was for 4.3 beta 1 or release candidate 1 or something, and I didn't see any report that it was fixed. My experience is with 4.1.1:
I installed Puppy on a friend's USB drive, and we played around with it on a computer at work. He then took it home and had to rerun the network wizard, and then brought it back to work and reran the network wizard, and then took it home and couldn't connect to the internet even after running the network wizard. The reason is that /etc/resolv.conf had somehow been replaced by a link to /etc/ppp/resolv.conf, which itself is a link to /etc/resolv.conf. On both computers we are connecting via ethernet to a router with DHCP. My friend is completely new to Linux so wasn't messing around with configuration files or anything, so somehow the gui wizard scripts or something must be at fault.

What I want to know is:
1. Does anyone have any idea why this problem would occur occasionally, or what is causing it?
2. Does anyone know if it has by any chance been fixed in 4.3?
3. Does anyone know a reason why the network setup wizard shouldn't delete the two resolv.conf files / links (or rename them with a .bak or something) to make sure they can be created properly?

In the Roaring Penguin pppoe package, I modified /usr/sbin/pppoe-setup back in July 2008:

    #v4.00 BK if enter 'server', somehow /etc/resolv.conf gets changed to a

    #symlink to /etc/ppp/resolv.conf (which is a symlink to /etc/resolv.conf)

    #so it's a circle. This script does not do that, something else does if it

    #sees /etc/resolv.conf is empty. A workaround is to put something into it,

    #what I have done is put in a free name server:

    if [ "$DNS1" = "server" ];then

     [ "`grep '^nameserver' /etc/resolv.conf`" = "" ] && echo 'nameserver 208.67.222.222' >> /etc/resolv.conf

    fi

I never did find out what actually causes the circular symlink problem.

However, my comment does indicate a possible solution, or at least a workaround. It all hinges on my comment "something else does if it sees /etc/resolv.conf is empty". If that is correct, then perhaps all I need to do is have a comment in resolv.conf, like this:

# nameservers go in here

I'm very nervous messing around with such a vital file at this stage, when 4.3.1 is imminent, but this possible fix is worth a try. Well, there will be a 4.3.1-RC, so give the network setup a good test, make sure that a pre-existing comment in resolv.conf doesn't break anything. I can't see why, but you never know.

Um, but maybe whatever causes that bug needs to see an actual "nameserver" entry in the file? We shall see.

Jwmconfig fix

A little while ago happypuppy reported a bug in the JWM Configure tool:

http://www.puppylinux.com/blog/?viewDetailed=01059

...which I passed off as not a bug, not really. Well, it isn't a bug in the strict sense, if you don't manually edit /root/.jwmrc-tray, only use the GUI tools.

Not very satisfactory though, and fortunately Patriot has taken a look at it and fixed it:

The attached file contains theme-switcher fixes for the zero-byte length jwmrc-tray issue. I have previously looked at the codes and as you have noted, the blinky and xload -bg update routines may render a zero-byte length jwmrc-tray after the sed ...

I've attempted to improve it a little bit with a more robust code. It should fix the zero-byte length jwmrc-tray issue. Please take a peek and see if any of the new codes can be used ...

Well, the attached code was in a p.m. to me, but I have put it into 4.3.1 which is coming RSN (Real Soon Now). The file is /usr/local/jwmconfig2/theme_switcher.

The PET package is now 'jwmconfig2-20091009.pet'.

430: 'man' fixed

Thanks to technosaurus for this clever fix:

I bashed together a fix for the broken man pages that restores 99% functionality
Code:

#! /bin/bash
defaulthtmlviewer "http://www.google.com/search?&q=man+"$1"+site:linux.die.net&btnI=Search"

I have put this into /usr/bin/man, specifically this at line 62:

exec defaulthtmlviewer "http://www.google.com/search?&q=man+\"${1}\"+site:linux.die.net&btnI=Search"

So, you can do something like:

# man e2image

...and yippee, it works!

We also need the same fix for /usr/share/doc/index.html, the main Help page. It has a Javascript function:

function search_site()

{

var url = "http://linux.die.net/man/1/"

// Append the text box input to the search url for the selected site

url = url + document.search_form.u_search.value;

// Change the the browser location to the new url

window.location.href = url;

}

...anyone know Javascript, can figure out how to put technosaurus's fix into that function?

430: GIT "broken"

ARAN reported:

The Package Git is broken especially "git pull" dont work !

I was just at the git irc chanell and asked what could be the problem and this was here the answer.

Quote:
hello to all. can somebody tell me how to resolve this ugly problem here ?/usr/libexec/git-core/git-pull: line 204: exec: git-merge: not found
git: 'pull' is not a git-command. See 'git --help'.

Did you mean this?
pull

this here are all files in git-core folder. could it be that something is missed ?
http://pastebin.com/m23ca0d9e

Quote:

Well, that error message really looks like some busted git installation.

Aw... Seems like they stripped out all direct executables for builtin commands, but didn't note that some scripts need the direct executables for those...

Yes, some files are missing out of that 'git-core' directory. I created a new PET pkg with all the files in 'git-core', but the PET is 49MB (yes I did strip), compared with 5.5MB before. I do not want to blow out the 'devx' by another 43.5MB!!!

I don't remember how I created my 5.5MB cut-down git pkg. Anyway, I have just put in 'git-merge', as that was the missing one in ARAN's post.

Comment: that sure is one bloated project!

430: frequency scaling

Aragon reported:

as pointed out by aran: frequency scaling is not working in 4.30.

this is a problem for pup-430-small.iso. small seems to miss more modules than i thought. having a scaling-app but no modules is a bug i think.

Ok, I have fixed that in the "small" build.

430: Aqualung?

Guys, what's the verdict on Aqualung in 4.3?

Personally, I don't like it. There was some feedback I think that it crashes.

Should I change to Pmusic?
Or, go back to how it was in pup 4.1x? (click on a sound file either launched a commandline player or Gxine).

Memtest in 2.6.30.5

In the 4.3 feedback thread one of the guys laments the absence of memtest.

I took it out as the kernel from 2.6.26 has a simplified memtest builtin:

http://kernelnewbies.org/Linux_2_6_26

If I boot with "puppy memtest loglevel=7", nothing is reported.

Looking at kernel git logs:

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=272b9cad6e7a2f61b13cfcd7dde0010e02e9376e

x86: early memtest to find bad ram



do simple memtest after init_memory_mapping



use find_e820_area_size to find all ram range that is not reserved.



and do some simple bits test to find some bad ram.



if find some bad ram, use reserve_early to exclude that range.

...which seems to indicate that bad ram is automatically reserved.

Looking in the kernel source documentation, kernel-parameters.txt:

memtest=	[KNL,X86] Enable memtest

			Format: 

			default : 0 

			Specifies the number of memtest passes to be

			performed. Each pass selects another test

			pattern from a given set of patterns. Memtest

			fills the memory with this pattern, validates

			memory contents and reserves bad memory

			regions that are detected.

...oh, just putting "memtest" as the boot parameter actually does nothing!!!?

So, I tried "puppy memtest=2 loglevel=7".

Still nothing. Run 'dmesg' after bootup, nothing in there. Look in /proc and /sys, can't find anything in there.

So, a memtest feature that reserves any bad ram it finds, but doesn't actually report any of its findings ...unless it only reports if bad ram found? ...I think that borders on useless -- well, no, not "useless", but less than satisfactory.

Note, the 2.6.30.5 kernel is configured with:
CONFIG_MEMTEST=y

So, for pup 4.4 we do need to consider putting memtest86+ in.

430: Shutdown problem when upgrade pupsave

This problem is reported in a couple of places, including here:

http://murga-linux.com/puppy/viewtopic.php?t=47169

I'm not clear on why this problem is occurring, anyway I have put this code into /etc/rc.d/rc.update:

  if [ $NEWPVERSION -gt 424 -a $OLDPVERSION -lt 424 ];then #424=4.3beta3

   if [ -d /usr/X11R7/bin ];then

    ln -snf /usr/bin/restartwm /usr/X11R7/bin/restartwm

    ln -snf /usr/bin/wmexit /usr/X11R7/bin/wmexit

    ln -snf /usr/bin/wmpoweroff /usr/X11R7/bin/wmpoweroff

    ln -snf /usr/bin/wmreboot /usr/X11R7/bin/wmreboot

   fi

  fi

The '-' versus '_' problem

I responded to alecz here:

http://puppylinux.com/blog/?viewDetailed=01129

...with a suggestion.

I have an idea. You have installed the files to a hd, you mount the hd from a running Puppy and the files look correct. But when you try and boot from the hd, Puppy cannot find 'pup-430.sfs'.

If that is your situation, try mounting the hd (while running Puppy) from the commandline using '-t msdos'. This is suggestion I replied to alecz:

Suggestion: mount the hd like this (if partition is sdb1):

# mount -t msdos /dev/sdb1 /mnt/sdb1
# ls /mnt/sdb1
# umount /mnt/sdb1

...if the names are wrong, let me know. You can correct them before unmounting.

Do try the above, not just the GUI tools in Puppy, as Puppy mounts a fat f.s with '-t vfat', and that might be hiding your problem.

Oh, and to provide me with more complete information, run this:

# disktype /dev/sdb1

...you can drag the mouse to highlight the text in the xterm window, then use middle-mouse-button to paste it anywhere.

Partview returns

I have been working through the 4.3-bugs forum thread, and a couple of guys have expressed sadness that the old 'Partview' application is not there.

Well, it was in Puppy 3.x, but gone in 4.0x and 4.1x as the required tcl/tk libraries were not there. However those libraries returned in 4.2x and so did Partview.

Once again in 4.3 Partview (and tcl/tk) is gone.

Actually all this time I was hoping that someone might write an equivalent using GTK, however noone did. So, today I decided to do it.

I knocked together a very simple Partview, that has both more and less features than the old one. This new one can have features added later if desired.

What it does not do is update, it is just a static display. An extra thing that it does do is show free space on all partitions, whether they are mounted or not. Snapshot:

...notice that it even shows free space in an optical disc, which is very handy if it is not closed, open for further burning, as in the case of multisession CD/DVD.

It is launched from the JWM tray, and as it is (currently) static, it has to be re-executed whenever you want to see the latest free spaces -- which is really ok, usually we don't want yet another app that has a polling loop taking up CPU time.

However, maybe could put a "refresh" button, or even auto-polling if desired. They are possibilities for the future.

Pwidgets in PPM

Pwidgets was missing from the PPM. There is a report that it works fine in 4.3, so I have added it to the PPM, in the 'puppy4' repo.

537 upgrade

Rerwin has posted upgraded '-3' Intel 537 modem udev rules and firmware:

http://murga-linux.com/puppy/viewtopic.php?t=46594&start=60

The new PET package is:

intel-537EP_secure-2.60.80.0-3.pet

Extra network modules for 2.6.30.5

Tempestuous has compiled extra drivers for the 2.6.30.5 kernel (as used in Puppy 4.3):

http://murga-linux.com/puppy/viewtopic.php?t=47267

As I am restricting 431 to essential bugfixes only, I am reluctant to put them in.

Perhaps many of them can go into 4.4?

PupDial upgrade

Rerwin posted this:

If a pupdial user checks the Stupid Mode box or unchecks the Auto Reconnect box for Account 1, those changes are not retained across dialup connections or for a subsequent pupdial session. This problem was introduced during 3G-wireless implementation and is easily fixed.

An ongoing concern regarding the wireless modem support has been that at least one ISP requires addition of a dialer command to permit connection. This forces the user to edit a file within the dialup structure, which should not be required of new users. In particular, VirginBroadband aborts a connection if the "chap" form of authentication is attempted. Forum member 01micko finds that adding "-chap" to the /etc/ppp/options file prevents the attempt and allows connection to that ISP.

I have added a simple, generic fix that adds the content of a file named for the ISP (matching the APN) to the wvdial files as it is copied to /etc/ppp/peers for each connection. The particular file resides in /etc/wvdial_options. Users who discover a need for a special dialer command can add an appropriate file to that directory. This implementation associates the use of the command to each account.

Thanks to 01micko's reporting of both issues, I attach a fixed pupdial and the new file for VirginBroadband, as well as difference files. Thanks, too, to gyro for raising the checkbox issue. I did not change the Reconnect default, but will wait to hear objections first and can update the attachment (to -2).

UPDATE 9/24/09: I have replaced the fix dotpet with better ones. Both dotpets have a more thorough implementation of the bug fix. Because the lines handled by that code are mainly entered by the user, the code ensures that the lines are entered only in the account sections of wvdial.conf and ignores such lines entered elsewhere in wvdial.conf.

The APN-specific option remains, except that the VirginBroadband entry is not to be included in puppy, instead being supplied by the user (who would know when the option may no longer be required). But the dotpet does create the file when the dotpet is installed.

To address purple_ghost's concern about Auto Reconnect being abused to keep a connection during long idle periods, I added support for a possible user-entered line: "Idle Seconds", which times out after a specified period of inactivity. I also added that item to the wvdial.conf help information accessible from pupdial. This allows protection from inadvertent wastage of connect time and cost, but also disables auto-reconnect. Setting this would prevent a customer from accumulating excessive connect times. However, the Auto Reconnect setting is not unchecked when reconnect is disabled. I am reluctant to change the default setting of Auto Reconnect, as I feel that is Barry's decision to make. But now turning it off (or having it disabled) yields the benefit of protection from runaway connect charges. (I originally hoped that "Idle Seconds" would leave auto-reconnect active, but that is not the case.)

The (-1) bugfix-only dotpet is provided in case Barry declines to accept anything beyond the serious bug fix for 4.3.1, although I hope he will consider taking the modest refinements, too.

He posted this PET package:

pupdial_update_to_P430-final-2-refinements.pet

...as well as an upgrade for the 'pupdial' script, a documentation file also had to be changed in the 'wvdial' PET package.

Sound and HSF-modem conflict

Rerwin posted this:

lwill and others report issues with sound cards and the HSF modem drivers. At least part of their problems involve the loading of HSF modem drivers instead of the sound-card modem driver. This is due to particular modems being supported by two or more drivers (in modules.alias). To resolve this, I have added to the PCI_OVERRIDES definition in /etc/rc.d/MODULESCONFIG entries forcing use of the sound-card modem drivers instead of the HSF counterparts.

In researching this issue, I discovered similar conflicts between some HCF and HSF drivers, so I forced one of them, based on their descriptions in the PCI_IDs listing. This confusion may be due to original misidentification of the HCF/HSF modems.

While updating MODULESCONFIG, I discovered that the a-f characters in the "hso" entries were erroneously capitalized.

The updated file is attached as a dotpet for anyone to try, along with difference files for it.
Richard

UPDATE 10/2/09: After reading that the modems that belong to both an HSF and ALSA driver probably need both, I have removed the overrides for them. iwill reports that the override did not prevent the HSF driver from running and that the modem did not work, although the sound card did. But the other changes are valid, for now.

And an upgrade for /etc/rc.d/MODULESCONFIG:

MODULESCONFIG_update_to_P430-2.pet

537 driver in 2.6.25.16 kernel

Rerwin posted this:

The kernel 2.6.25.16 version of Puppy 4.3 does not detect these modems, because the modem-rules file and firmware tarball are the incorrect versions. I obtained the appropriate versions from puppy 4.1.5. The attached dotpet applies ONLY to the k2.6.25.16 puppy installations. The two files should be added to the bugfix release 4.3.1-k2.6.25.16.

And he posted this package:

modem_fix_to_P430-k2.6.25.16-1.pet

I have applied this fix to the 2.6.25.16 kernel.

Intel 537 modem drivers fixed

Rerwin has fixed a major bug in the 537* drivers that caused a bit of a panic when I uploaded 4.3-final -- had to remove those modules and re-upload the 4.3 iso.

He posted these packages to the forum:

intel-537EP_secure-2.60.80.0-2.pet
intel-537EP_secure-2.60.80.0-k2.6.30.5-intel537-2.pet
intel-537EP_secure-2.60.80.0-k2.6.30.5-intel537EA-2.pet
intel-537EP_secure-2.60.80.0-k2.6.30.5-intel537EP-2.pet
intel-537EP_secure-2.60.80.0-k2.6.30.5-intel537SP-2.pet

Which have the fixed drivers and updated firmware.

Note to rerwin: I don't yet have the latest source for the 537 drivers. Could you post it so I can upload it to puppylinux.com/sources?

Remaster CD bug fixes

There was a report in the 4.3 bugs thread on the forum, of the CD remaster script crashing. I couldn't reproduce that, but I did find 3 other bugs.

1. Boot logo and help missing.
2. mksquashfs error that may have caused iso to be bigger.
3. Hiccup in manipulating the database files in /root/.packages that may have compromised operation of the PPM.

There was a request that the remaster script needs a choice where to get the Puppy files from. Currently they must be on a CD. Need to also be able to choose a mounted directory.
However, at this stage I am being very careful, essential bugfixes only for 431.

One of the bugs in 430-final was caused by an earlier beta-release bugfix. So, I need to be extremely careful.

Sylpheed 2.7.1

Prehistoric reported that Sylpheed (PET package from ibiblio, v2.4.7) crashes under certain circumstances.

I have compiled the latest, version 2.7.1.

This is not in the standard live-CD but will be a PET package at ibiblio.

WARNING TO WINDOWS USERS

In the 4.3 bugs thread in the forum, there are a few reports of the files on the Puppy 4.3 live-CD being incorrectly named. Especially "PUP_430.SFS".

What needs to be understood is that this is not a new "bug", all prior puppies have it.

The reason that you are seeing the wrong filenames is because you are using MS Windows. That is, running Windows, you plug in the Puppy CD and view the files in Windows Explorer.

I presume that you are doing this to copy the files for doing a frugal hard-drive install or USB Flash drive install.

The Puppy live-CD ISO file (CD image) is created without Joliet extensions -- this is needed for MS Windows to read the filenames on a CD/DVD correctly. This is what causes the problem.

The reason that Joliet extensions is left off is because I had experience that it breaks saving of sessions for the multisession-CD/DVD (saving sessions back to the CD/DVD, no hard-drive or other storage media required on the PC).

So, the question must be raised, why are you running Windows to copy the files off the Puppy CD? You need to wean yourself off this dependence on Windows. Boot the live-CD, then you have a running Puppy!

Alternatively, maybe you only have the downloaded 'pup-430.iso' live-CD image, you have not burnt it to CD, and you are running MS Windows, and you use a Windows application (Isobuster?) to view the contents of the ISO file and copy out the files.
Really, you "should" be running Puppy or some other Linux, but if you really must be running Windows to perform this operation, then copy the files and then rename them...

This is how Windows sees the files:
BOOT.CAT BOOT.MSG HELP.MSG INITRD.GZ ISOLINUX.BIN ISOLINUX.CFG LOGO.16 PUP_430.SFS VMLINUZ ZP430305.SFS

For frugal install, copy the required file to the destination then rename them to:
initrd.gz pup-430.sfs vmlinuz zp430305.sfs

...most important, rename PUP_430.SFS to pup-430.sfs!

Note, viewing inside an ISO file when running Puppy is a piece of cake. Just click on it!

430: Mouse Wizard fixed

The Mouse Wizard showed "USB mouse" selected when the hardware was actually a PS/2 mouse.

Puppy no longer distinguishes between the two. That is, it is now taken care of by the kernel driver.

So, I have merged the radiobutton in the Wizard, now just two radiobuttons "Serial mouse" and "USB or PS/2 mouse (also touchpad)".

430: PPM configure fixed

This is a continuation of the previous two posts, fixing the Puppy Package Manager.

In the 'Configure' section of the PPM, clicking on 'Update databases' downloads them from ibiblio.

However, the list of repositories in the Configure window grows longer. Fixed.

430: Ensure dependencies installed

This is a follow-on from the previous post, debugging the Puppy Package Manager.

If I choose to install Sylpheed, the first window identifies 'bogofilter' as a dependency. There are two buttons "Examine dependencies" and "Install sylpheed-2.4.7".

What is not clear is that if the latter is clicked, it only installs Sylpheed, not any deps.

I have added clarification that "Examine dependencies" must be clicked to install Sylpheed as well as the dependencies.

Note, even if Sylpheed only is installed, the final post-install window will identify what packages are missing.

430: Package check dependencies broken

The 'Setup' menu, item 'Check dependencies installed pkg', was broken.

I am testing with the Sylpheed package, it has the dependency 'bogofilter'. I installed without the dependency and found this problem.

The problem also occurs in the main PPM. If I choose Sylpheed, a window identifies the dependency 'bogofilter' is required. I then go ahead and install Sylpheed only, but the final window states that there are no missing dependencies. This is also fixed.

Verified Open Office SFS works

Ramachandran reported:

I have downloaded the open office sfs file put it mnt/home. However, the same is not working. On clicking the SFS file it shows mounting. But nothing happens. I have tried system-boot manager.However, it showing no sfs in mnt/home ! What to do

There are a couple of other similar reports in the forum 431-bugs thread.

I have just been through the whole exercise. A pristine 430 running from CD, downloaded Open Office 3.1.1 SFS from ibiblio, to /mnt/home.
I chose it in the BootManager, rebooted, it works fine.

It works.
Please check the md5sum.
Please download a SFS version 4 if running the 2.6.30.5 kernel, or SFS version 3 for older kernels.

It is difficult to go wrong. If you have checked the md5sum, then the file is ok. When you run the BootManager it pops up a window telling you if any SFS files at /mnt/home are the wrong version.

So, it works.

Er, you do have a "pupsave" file and a partition mounted at /mnt/home?

430: Menu not update when package installed

Prehistoric reported:

I just found a clue about the missing menu entries on pet packages installed in 4.3 with the latest PPM. When I installed gimp 2.4.0rc3.pet downloaded directly from ibiblio there was no menu entry. When I clicked on a local file of the same pet there was no report it was already installed, and the menu entry appeared as expected. In the previous case, where there was no menu entry for Sylpheed, I had also downloaded from ibiblio using the PPM interface. Downloading and installing by clicking on a local pet file did solve the missing menu entry problem.

Yes. The script /usr/local/petget/petget is what runs when you click on a downloaded package. When you run the Puppy Package Manager and choose a package from a repo (and a local directory can be chosen instead of a URL), then /usr/local/petget/installpreview.sh runs, not 'petget'.

Both of these have code to call the script that updates the menu, however there was a bug in 'installpreview.sh' that prevented the menu update. Fixed.

Release schedule for 431

Technosaurus is working on 4.4, however it would be nice if 431 was available, then 4.4 development could take off with that as the base.

So, it is probably a good idea if I announced a release date for 431, or at least something to aim for.

It really depends on how many bugs I want to fix. 431 is supposed to be a bugfix release of 430, so should not have any other changes.

I was thinking of setting the release date for the 1st of November, for the reason that my satellite Internet account resets it's quota at the start of each calendar month.

I'm going to be testing Woof soon, and this is going to chew through my October allowance (4GB).

As the so-far and anticipated bugfixes for 431 are pretty straightforward, probably only one or two pre-releases will suffice.

I'll aim to get one of those out mid-October.

Does that schedule seem ok? Too slow?

Spanish locale fix

Forum member myselfo reported this:

In the new locale selection screen during the bootup when no previous session is saved, there is something strange:

an-ES Spain

But that's not correct, the code for Spain should be es-ES... but when I try to select es-ES it appears as "Mexican, Spain". Actually the language is listed as Mexican for every Spanish-language country, which is kind of funny, but wrong Smile
It should be "Spanish, Spain", "Spanish, Mexico", "Spanish, Chile", and so on.

I fixed the script /usr/sbin/chooselocale.

But, what is the "an" part of "an_ES"?

...ah, found it. The "an" is "Aragonese", which is a minor language in Spain.

I fixed a couple of others too, such as "el_" is "Greek" not "Cypriotic". "mk_" is not "Fyrom", it's "Macedonian" (?).

HOWTO satellite videos

Coolpup sent me a link to this great site, has videos that explain all aspects of installing satellite TV:

http://www.ehow.com/videos-on_10573_satellite-tv-installation.html

Computer cartoons

These come from here....
http://stackoverflow.com/questions/84556/whats-your-favorite-programmer-cartoon

I'm wasting lots of time, better get back onto Puppy bug-finding.

Satellite TV

I'm planning to install a satellite TV system, as currently I have a terrestrial TV antenna on a high mast that can only receive one channel, ABC1, and that with (sometimes) considerable snow.

Here in Australia, Optus has two satellites with free channels, C1 (has ABC1 and some commercial channels) and D1 (has ABC2) -- there is more, such as SBS and some foreign channels. They are on the ku-band.

The cheapest supplier I can find in Western Australia is these guys (AU$295):

http://www.sciteq.com.au/store/product_info.php?cPath=26_27&products_id=163

To receive from both satellites, I need two dishes (90cm dishes cost $55 each) and a cable-coupler thingy called a "DiSeqc/tone burst switch".

Although these channels are "free", I do have to purchase a smartcard, included in the above package, but AU$85 separately (actually, I priced all the items in the above package, and it adds up to less than the package price!).

I think that I understand basically what all the parts are for -- anyway, they provide an installation guide.

I'll also buy a few extras, like roof bracket mounts, cable, compass ($10), signal-finder ($15).

My little house with 3 dishes on top is going to look very hi-tech!

Anyone reading this who has done an installation? Any tricky things to watch out for? -- yeah, well I suppose, getting the dish pointed right at the satellite is the hardest part!

Optus D1:

4.3 Japanese edition beta1

Snapshot:

Announcement:
http://murga-linux.com/puppy/viewtopic.php?t=46704&start=15

Tsunamis and earthquakes

There have been recent posts to this blog and the forum about the typhoon in the Philippines.

Recently, like in the last few days, there have been earthquakes and a tsunami around Australia, with major damage and lost of life. The vast Australian tectonic plate is on the move. A plate in the Pacific is pushing underneath the Aussie plate, and at the other end the Aussie plate is going underneath a Eurasian continental plate. At least, that is my uneducated overview after watching the news last night.

http://www.kval.com/news/local/62983817.html
http://www.watoday.com.au/world/second-major-earthquake-hits-sumatra-as-death-toll-soars-20091001-gd48.html

Interesting Google map:
http://www.google.com/maps/ms?ie=UTF8&hl=en&msa=0&msid=117631292961056724014.000474d161177b2aa613d&t=h&ll=-20.159572,144.140625&spn=45.3039,101.601562&source=embed

The China Study

This post is definitely in the "General" category, not "Woof"!

There is a thread running in the forum about the Swine Flu vaccine, would you take it, to which I posted a response. Then while on my morning walk my mind went off into a whole lot of memories about my own and others medical experiences. Somehow I got onto thinking about diet -- well of course our wellbeing is strongly related to diet.

I'm a vegetarian, have been since ...um, over 33 years. I'm what is known as a ovo-lacto-vegetarian, meaning I take diary products and eggs. A couple of years ago I was most surprised when my daughter became a vegan -- I had previously considered them to be the lunatic fringe.

The thing is, a lot of what we take to be as "fact", turns out later to be bogus. The same holds for medical professionals -- after a lifetime of experience, my opinion of them has been steadily dropping. I think that generally they are good at "physical" things, like cutting out a tumour or stitching up a cut. But beyond that, many of them, especially GPs, are quite ignorant.

For the first year that my daughter became a vegan, she went to a doctor for regular checkups. She also had her blood checked. On one visit to get her blood test results, her blood was excellent, her general health good, but the doctor persisted in being very critical of her being a vegan. My daughter retorted "In your 5 year degree you study nutrition for just 5 weeks, and you think that qualifies you to lecture to me as though you know it all!" Yeah, well, the doctor wasn't happy. That's my daughter, she has turned out to be a very strong character.

As is typical of the "newly converted", she often prompts me to adopt veganism. Like a couple of weeks ago, I mentioned that I still eat eggs, and she responds "you know what they do to the hens after they stop laying, don't you!". Sigh, yes, I can imagine.

Anyway, there's a book that she got me to read, "The China Study". It's about as anti-quack as you can get, very heavy reading, but I did read it right through. I still didn't become a vegan, but I found it fascinating for various reasons -- one of which is how our knowledge about medicine and diet, what we consider as "fact", is highly manipulated by vested interests.

The principle author was a very high-profile researcher who was in charge or co-charge of many major studies, and served on many government panels. His stories of behind-the-scenes twisting and hiding of the truth are fascinating.

Personally, I don't think that you "must" become vegetarian or vegan. The essential thing that the book concludes is the importance of a natural "whole food" diet that is high in plant content, low in animal content.

I think that it is one of those "must read" books.

http://en.wikipedia.org/wiki/The_China_Study
http://www.thechinastudy.com/

Finally, for the sake of balance, there's this very critical review of "The China Study":

http://www.cholesterol-and-health.com/China-Study.html

...which also has a link to a response from Dr Campbell.

Boot and save ext4 partition fixed

Thanks to help from forum members piratesmack and tulindo, this is now fixed. Forum thread where they posted:

http://murga-linux.com/puppy/viewtopic.php?t=46902

Thanks very much guys. I'm very pleased when you guys get stuck in and find the solution to a problem. It sure does make things easier for me.

Note to tulindo: there are two places in rc.shutdown where "ext4" has to be inserted.

These fixes will of course be in the upcoming 431.

Guide for Woof

Forum member Iguleder has written a guide to using Woof:

http://murga-linux.com/puppy/viewtopic.php?t=47331

I'm writing a GUI for Woof, but the commandline scripts will still be there, and many users will probably continue to use them. So, the guide in its current form will remain relevant.

Note, the GUI is well advanced. I intend to release Woof with GUI at the same time as 431.

430: Insert key crash fix

Thanks to pizzasgood for this, I never would have figured it out.

Under certain conditions, not always, pressing the key while a Mozilla-based browser is running, causes it to crash. The test I did was to start PPLOG, which uses MU's web browser, start a new post and then press the key while the cursor is in the text-entry box -- result, crash.

Pizzasgood narrowed the problem down to one library, libgdk-x11-2.0.so*, and found that the gtk+ package had to be configured with '--enable-debug'.

I looked at what ttuuxxx did for his bug-fixed "431.1" build. He compiled a later version of the gtk+ package, 2.16.x and reported that it spits out error messages, although it does work.

I firstly followed ttuuxxx's lead, compiled gtk+ 2.16.6 and glib 2.19.10. I compiled these with '--enable-debug=no', and found still had the crash bug.

Then I went back to the same gtk+ version as in 430, 2.14.7. I recompiled it with this configuration:

./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --build=i486-pc-linux-gnu --with-xinput=yes --enable-debug=minimum

Great, PPLOG did not crash!

I have upgraded the gtk+ 2.14.7 PET package with the new library files:

libgailutil.so.18.0.1
libgdk_pixbuf-2.0.so.0.1400.7
libgdk_pixbuf_xlib-2.0.so.0.1400.7
libgdk-x11-2.0.so.0.1400.7

I did not upgrade libgtk-x11-2.0.so.0.1400.7 as the new one is 4.5MB compared with 3.5MB before.

The new PET packages are named:

gtk+-2.14.7-1-p4.pet
gtk+_DEV-2.14.7-1-p4.pet

...I'll upload them soon.

I'm back home

My Royal Show job runs until Saturday, but I have experienced back problems so left. Fortunately they had someone on standby who could take my place.

When I got up one morning and could hardly stand up, and couldn't bend over to put on my boots ...I knew that I had a problem!

Lots to catch up on....

Technosaurus has volunteered to coordinate 4.4. Wiki page, plus other guys willing to help:
http://puppylinux.org/wikka/Puppy44Team

Technosaurus has been asking for suggestions for 4.4CE:
http://murga-linux.com/puppy/viewtopic.php?t=47065

Technosaurus has started "phase 1" of testing for 4.4CE:
http://murga-linux.com/puppy/viewtopic.php?t=46941

Technosaurus, thanks for stepping forward to tackle this. You'll get a buzz out of it. And the experience of being the focus of lots of people clamouring for what they want in the next Puppy!

OpenOffice, Anjuta SFSs on ibiblio.org

I have uploaded Open Office 3.1.1 and Wosh's Anjuta 2.4.2 to here:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/sfs4_modules-4/

The Open Office SFS has Java, and if I've done it right the web browser Java-plugin should also be functional.

I posted about Wosh's Anjuta SFS here:

http://puppylinux.com/blog/?viewDetailed=01109

Note that the 'devx' SFS file is also required if you want to use Anjuta:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/puppy-4.3/devx_430.sfs

These links are all to "sfs4" (Squashfs version 4.0) files, for use with the 2.6.30.5 kernel.

Anjuta IDE for 4.3

Great news, forum member Wosh has created an SFS for the Anjuta C/C++ IDE, specifically for Puppy 4.3. Announcement and download here:

http://murga-linux.com/puppy/viewtopic.php?t=28629&start=60

I haven't used Anjuta myself, but it sure does look good. Anjuta home:

http://projects.gnome.org/anjuta/index.shtml
http://projects.gnome.org/anjuta/features.shtml

Note, sometime ago there was talk about offering Vala support in Anjuta, but I don't know the current status on that.

PUP_430.SFS "bug"

Forum member Rickrandom reported:

In Windows (oh the shame, I guess Puppy will look inside an ISO without burning to a CD, but I don't know how) pup-430-small.iso contains PUP_430.SFS

Yes, shameful

The iso was created without Joliet extensions. It has been done this way for many previous versions of Puppy also. Joliet extensions causes trouble -- straining my memory but I think it caused the multisession saving to fail.

You should be using Puppy. Puppy does everything.

Why are you stating can't look inside an ISO? Of course you can: just click on it.

Developing a GUI for Woof

I have started writing a GUI frontend for Woof.

This will take Woof to a new level of ease-of-use. One thing it should do is make it easier to add or remove packages and repositories.

Back home 4th Oct

Just to let everyone know.

I can't interact much on the Internet, as away from home. Working hard on other stuff, just doing a tiny bit of Puppy stuff.

I'll be home on the 4th October, will catch up with everything then, eager to get stuck into Puppy development again.

Have several CD orders, will post them tomorrow.

Open Office, soon

Regarding the requests for Open Office SFS, I haven't had Internet access as away from home.

I'll see if I can do it tomorrow.

Uevent, SFS fixes

I am not able to get much Internet access. Right now I'm visiting one of my sisters to use her internet.

I have done a little bit of work on Puppy...

I already reported a problem with hardware detection with the 2.6.21.7 kernel. This is due to the way the kernel handles uevents in kernels prior to 2.6.24.
The result is that some modules may not load, though I expect it to be ok for most people.
Anyway, I fixed it, a simple modification to /etc/rc.d/rc.sysinit.

Another problem is that the desktop icons can get messed up when SFS files are added or removed.
This is also fixed, needed a change in /etc/rc.d/rc.update.

Puppy CD

I have created an new CD for Puppy 4.3, the one I sell for $9.

It has the same as 'pup-430.iso', that is, all the drivers and kernel 2.6.30.5.
It also has Woof, and the PET packages required to build Puppy using Woof.

Also I managed to squeeze in the 'devx', the kernel source, and Open Office 3.1.1 SFS files.

Total size is 695MB.

There are a few orders to process, I'll post them on Monday.

Oh yeah, I created an SFS 4.0 of Open Office 3.1.1, with JRE.
I looked at ttuuxxx's OO PET package, but the JRE didn't work, so I downloaded JRE and got it to work.

I will upload the Open Office SFS in a day or two.

pup-430-small.iso

There seem to be some satisfied customers using the "small" live-CD. If 96MB can be considered small. The full build is 105MB.

The small build is supposed to hve enough drivers to satisfy "95%" of users. If there is some really important driver missing, that is in the full build, let me know.

I don't want the small build to creep up in size though, adding less-commonly-required modules, I would rather identify more that can be taken out.

So the reverse applies, if you can see a module or group of modules that you reckon 95% of users will not need, let me know.

I would like to see if I can drive that 96MB even lower.

'split-pup' re-uploaded

I have re-uploaded the split files:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/puppy-4.3/split-pup/

Ha ha, toward the end, my Internet speed got "shaped" to dialup speed. I have used up my monthly quota -- 2GB peak, 2GB off-peak.

Interesting Skymesh's idea of dialup speed -- the upload to ibiblio was a steady 4.6KB/sec.

I get my next allotment at the start of the next calendar month. No problem, I'll be working in Perth for the rest of the month. This time I'm staying with a relative who doesn't even have a computer!

ttuuxxx's Puppy "431"

Ttuuxxx has been remarkably fast!

He has already created an upgrade for 4.3:

http://murga-linux.com/puppy/viewtopic.php?t=46727

He has built it with Firefox instead of SeaMonkey.

Note that removing SeaMonkey has a flow-on effect, rendering some apps unusable, such as PuppyBrowser.
Ttuuxx has posted a patch that is supposed to fix that.

I haven't tried it. It's 2.30am here, I'm going back to bed.

Kernel PET re-uploaded

For you guys who are playing with Woof, I have re-uploaded the 2.6.30.5 PET package:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/pet_packages-woof/linux_kernel-2.6.30.5-tickless_smp-5-p4.pet

I left the package name the same, so that Woof itself does not have to be changed. Just download this latest and replace the previous one.

The only difference is the Intel 537 modules are removed.

pup-430.iso re-uploaded

Here I am, out of bed at 1.15am again.

'pup-430.iso' has finished uploading.

I decided a new strategy is needed for uploading, so I named the file pup-430.isoTEMP then uploaded it, then named it back.

Yeah Sage, the iso is now 105.1MB. That's how big the 537 modules were -- added 5MB onto the iso!

I'm currently re-uploading the split files, and haven't yet checked the md5sum of 'pup-430.iso'. It should be:

e18650d38c4185a4b70778026c305e98

depmod has gone mad

Puppy 4.3 is using the Busybox 'depmod' applet because the full depmod uses too many resources.

However, the ALSA Sound Wizard needs the full depmod to work. Note, the full version is named 'depmod-FULL'.

But, depmod-FULL has gone crazy. It worked right up until a day or to ago. I modified the script /usr/sbin/alsaconf to run the full depmod, tested it, all was fine.

I have no idea why it has gone crazy. What it does is create /lib/modules/2.6.30.5/modules.pcimap.temp that just keeps growing bigger and bigger until there is no memory left.

Hmmm, looks like we will need a service pack to fix this one.

Anyone want to have a go at compiling the latest module-init-tools package, see if they have fixed this problem?

Though, what the cause is, has me stumped. I wonder if one of the 3rd-party modules is causing the problem?

EDIT:
I had tested the ALSA Wizard very soon before releasing the final, so I thought hard, what changed before then and the final? ...ah, one thing, I put rerwin's 537 modem modules in. That's it, take them out and depmod-FULL works.

You will find them at /lib/modules/2.6.30.5/intelmodem

This bug does not affect pup-430-small.iso, only the full size pup-430.iso.
I'm going to rebuild and re-upload it, without the 537 modules.

Puppy Linux version 4.3 released

Please note that the files have not yet fully propagated from ibiblio.org to the mirrors. For example, I just now checked 'nluug.nl' and it has an incomplete 'pup-430.iso' file. So, for now only download from ibiblio:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/puppy-4.3/

Announcement and release notes:

http://puppylinux.com/download/release-4.3.htm

There are a selection of live-CD ISO files and support files to choose from. Please read this page before going to the download page:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/puppy-4.3/readme-files.htm

Main download page, has links to ibiblio.org and download mirrors:

http://puppylinux.com/download/index.html

A thread has been created on the Puppy Linux forum for bug reports and suggestions:

http://murga-linux.com/puppy/viewtopic.php?t=46594

4.3: not there yet

Still uploading. Right now it's 1.30am, I have just got up and recommenced uploading.

My satellite connection died at 9.30pm so I went to bed.

That's the downside of satellite Internet, very unreliable. At least, that has been my experience.

It was the same story with my previous satellite provider. With the previous provider, the sync with the satellite would drop out frequently.
With my current provider there appears to still be a connction with the satellite, judging from the lights on the satellite-modem, but nothing happens.

When "nothing happens", and I'm using gFTP to do a transfer, gFTP just sits there indefinitely, never times out, nothing, just waits. I find it best to just power off and try again later.

Anyway, here we are at 1.36am and it's working.

I got a warning from my provider that my quota is just about used up -- then I get shaped to dialup speed. But, I think there's a bit more available on the off-peak quota, which is what I'm using now.

Enough rambling.

Uevent bug in old kernel

Ah, I have found a bug running 4.3 with the old 2.6.21.7 kernel. I won't let it hold up the release of 4.3 though.

In /tmp/bootsysinit.log there are several lines like this:

grep: /sys/bus/usb/devices/usb5/uevent: Permission denied

This is due to how I read the modaliases in /etc/rc.d/rc.sysinit. The 2.6.30.5 kernel allows 'uevent' files to be read and the MODALIAS variable extracted from them. I don't know what kernel version that came in, but 2.6.21.7 doesn't support it.

With the 2.6.21.7 kernel there are actual 'modalias' file in /sys that have to be read.

rc.sysinit loads modules in two stages, and it is the second stage that is now failing. This means that some modules might not load. There are not too many cases that are caught by this secondary loading step, maybe some PCMCIA interfaces and some less-usual USB interfaces.

So it's not a critical thing, and I can easily provide a patched rc.sysinit soon to fix it. It needs a kernel-version test, with code to do it the "old way".

Probably those who want to run 4.3 with this old kernel will discover more things, and we can easily provide a "service pack" prior to 4.3.1.

But, I must emphasize that the bug will only affect a limited range of hardware. I rebooted my PC, with a savefile, it came up, everything still working, all modules loaded.

4.3 with 2.6.21.7 kernel

Amazing! I'm posting right from Puppy 4.3 built with the old 2.6.21.7 kernel.

This kernel is about two years old. It was used back in Puppy 3.x and also retro 4.1x and 4.2x builds.

The thing is, I just built the Puppy live-CD with exactly the same packages as the main 4.3 build with 2.6.30.5 kernel. Some packages are supposed to be kernel-version sensitive. The kernel ALSA drivers are much older than the alsa-lib and alsa-utils packages.

But, everything works. I'm online using it right now, via wireless with Ndiswrapper.

I even had a bit of a pleasant surprise with the acx_pci driver. This PC has a Netgear WG311 wireless card, but I never was able to get the acx_pci driver to work with it. Now, it connects, DHCP works, but there is still a problem, /var/log/messages reports frame sizes to big.

There's probably an upgraded driver that would fix that, but I used Ndiswrapper and that works fine.

I also tested dialup on my other PC with Pctel 789 modem, that works too.

Media players work, everything works...

Oh, some news, 4.3 is going to be available with 2.6.30.5, 2.6.25.16 and 2.6.21.7 kernels.
They are uploading now, but you will have to wait several more hours due to ibiblio.org low transfer speed.

I'll announce when everything's ready.

Network Wizard upgrade

Tempestuous provided updated and new firmware for the 2.6.30.5 kernel, then discovered that the Network Wizard also needs to be upgraded.

4.3beta3 has Dougal's Network Wizard dated 4th April 2009. I have inserted the following line into /usr/sbin/wag-profiles.sh, at line 322:

ar9170usb|at76c50x-usb|libertas_cs|libertas_sdio|libertas_tf_usb|mwl8k|usb8xxx) CARD_WPA_DRV="wext" ;; #v430

I have named the upgraded Wizard as 'net_setup-20090916.pet'.

MSDOS expert required

It's too long ago, I can't remember...

When 4.3 is released, I will be uploading it in pieces, to make it easy for those on dialup.

So, pup-430.iso will be split into xaa, xab, xac, xad, xae, etc.

After downloading, if running Linux it is easy to put them together:

# cat x* > pup-430.iso

However, does anyone know the equivalent to that in a Windows DOS-box?

Um, I presume that you can still open a DOS-box in Windows? Perhaps the average Windows user would need instructions how to do that, open it wherever they have downloaded the files?

I'll provide the pieces on ibiblio, and a readme that tells how to assemble them.

Of course, in the future they will be able to use our Xdelta file difference manager.

Dialup works!

I'm doing a series of basic sanity checks on 4.3.

I went through the exercise of a frugal install, then tested dialup -- yes, working!

The modem in this PC has a PcTel 789 chipset, so it's a software modem. Piece of cake to run PupDial and connect.

I have signed up for a full year with Netbay Internet, the people who provide my dialup.
Although I have satellite broadband, my dialup account is a backup and also I need it for testing purposes.

Universal Installer bugfix

I did a full hd install of 4.3 and encountered a bug. The desktop came up with mostly the Rox default icons, not the current theme icons.

The particular circumstance that caused this was I previously had 430 installed on the partition but chose to wipe when testing the latest pup. The installer runs /etc/rc.d/rc.update in the newly installed partition, and it is supposed to perform some actions based on it being a version upgrade -- however the installer had already recorded that version 430 was in the partition so did not know that upgrade actions were required.

Well, anyway, the end result was that rc.update did not run the code that installs the icon theme into /usr/local/lib/X11/pixmaps.

Fixed.

4.3 frozen

Consider 4.3 to now be "frozen". I'm going to do a build then test as thoroughly as possible for a day or two.

There are a few things that I would have liked to put in, but will consider them for 4.3.1. For example:

USB floppy formating (8-bit)
Woo-FF video file converter (trio)

Updated firmware for 2.6.30.5

Thanks to tempestuous for finding the latest firmware for wireless drivers in the 2.6.30.5 kernel.

I have upgraded the 2.6.30.5 package with these.

NicoEdit 2.3

Nicolas has done it again.

This remarkable little text editor now remembers the previous scroll position in a file and the cursor position. So, in the History menu you can flip back through files and they all come up in the same place at which you last had them open.

This is great, makes NicoEdit just as good as, perhaps even better, than an editor with tabs.
Better perhaps, as you can have a large number of files in the history menu, whereas the tabbed interface gets cumbersome when you have too many to fit in the width of the window.

Anyway, as I am now at the 12th hour for 4.3, I have left Geany in also. We do need to use NicoEdit for awhile, build up confidence in it, maybe fix any outstanding bugs, and maybe consider NicoEdit as the sole GUI text editor for 4.3.1.

GTK Polished Blue theme tweaked

The guys were discussing on the forum that it is difficult to determine which tab is active in SeaMonkey, as the active tab is lighter grey, the reverse of what is expected in some other themes.

I responded that we could look into tweaking the theme for pup 4.3.1, but afterward I thought why not do it now.

So I have. I have made the active tab a light 3D blue, to match the other blue highlighting in the GTK theme and in JWM.
It looks nice and it is now crystal clear which tab is selected.

Ayttm 0.6.0

I have been using snapshots from CVS for awhile, the most recent was 0.5.0-146. There were some small changes made since then and the new official release is now version 0.6.0. So, I have upgraded.

A note while I think of it... technosaurus has brought out a PET package of 0.6.0 but I think it is rather large. I don't know why, mine is 806KB, with all the modules. I mention this as it is something for technosaurus to look into why that is so... maybe an extra pkg bundled with it?

puppyPDF fixed

See the 'Document' menu. This was created in 2007 by forum member thoughtjourney, and hasn't been changed since.

Until now, that is. Forum member nikin has fixed a couple of bugs:

I tryed to use puppypdf today and it had 2 showstopping bugs.

First: the dialogs where communicating on stderr, so they catched the GTK errors. messing up fselect boxe's filenames.



Modified them to use --stdout and 2>/dev/null



The other one was a misuse of abiword.

not working:



Code:

abiword --print=temp input





i updated it to work with abiword 2.6



like:

Code:



abiword --to=ps --to-name=temp input





this version now converted my html file to pdf

Any suggestions are welcome.

'zdrv' bugfix

4.3beta3 has a 'zdrv' of extra modem drivers.

When this is placed at /mnt/home, it loads automatically and 'depmod' runs so that the modules will be recognised. Ditto when the zdrv is removed.

At least it is supposed to. I found that 'depmod' did not run. Fixed, I think, still need to do a build to verify.

mhWaveEdit 1.4.16

4.3beta3 testers are invited to test the latest mhWaveEdit:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/pet_packages-4/mhwaveedit-1.4.16-1-p4.pet

And Zmixer, see my previous post.

In particular, are you able to record with mhWaveEdit? Lobster likes to record stuff, so it's a pity that recording does not seem seem to work on his hardware with ALSA 1.0.20 -- I hope that is not the common experience!

The 4.3beta3 forum feedback thread has some reports of audio problems, so we will probably need to upgrade the ALSA drivers for Puppy 4.4.
Probably stay with the same kernel version perhaps, Rerwin (and myself to a much lesser extent) has been through a long process of getting 3rd party drivers to compile with it.
Moving up to the latest in the 2.6.30.x series would be a reasonable thing to do. But then, you have to consider that there may be other 3rd party drivers compiled for 2.6.30.5, maybe by tempestuous, so there would be a big investment in staying with 2.6.30.5.
...but, this is a decision that the 4.4 coordinator would have to make.

Zmixer

It is approaching the 12th hour and I should not be adding any more packages, but I think that I will add zmixer.

AlsaMixer is a console ncurses application that is not to eveyone's liking, whereas Zmixer is a GTK2 app. I will have both in 4.3.

4.3beta3 testers, those with working sound, are invited to test Zmixer:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/pet_packages-4/zmixer-0.2.1.pet

Sweep vs mhWaveEdit

Charlie6 asked the question, how to record in mhWaveEdit.

I have taken a quick look, it doesn't seem very intuitive or flexible.

Maybe I should change to mhWaveEdit?

Nathan Wallpaper Setter fix

The 'View' button was broken. Thanks to 01micko who found the cause of the bug.

I changed it to 'defaultimageviewer'.

pupX fixed

Thanks to DaveS who advised that the '-dpms' parameter is required, I have fixed pupX. The bug reported was that disabling the screensaver did not work.

This fix is also required in /usr/bin/gxineshell, a wrapper script for Gxine, that turns of screensaving while a video is playing.

PupDial, ESS modem driver

Rerwin posted an upgrade for PupDial and a driver for ESS modem chipsets, which just missed out on getting into 4.3beta3.

I have now put them in.

The inclusion of ESS makes pup 4.3 with 2.6.30.5 kernel the most complete selection of analog modem drivers that we have ever had.

The only slight inconvenience is that they are in a separate 'zdrv' (due to their size), that the user will have to download and place at /mnt/home.
On the plus side though, a 'zdrv' is loaded automatically and does not have to be selected in the BootManager.

I have documented the extra modem drivers 'zdrv' in the 4.3 announcement page, with a list of what modem chipsets are likely to need it.

There are some software-modem drivers that have remained builtin to the live-CD iso file, such as Lucent and Smartlink.

NicoEdit 2.1

I announced NicoEdit 2.0 here:

http://puppylinux.com/blog/?viewDetailed=01049

But I later went back to Geany, plus NicoEdit 1.6 as the secondary "Leafpad replacement", as 2.0 has some bugs.

Nicolas has fixed the bugs in 2.0, and once again I am making NicoEdit the flagship text editor for 4.3.

I have done some limited testing, it looks good. As I already mentioned, it is tiny -- the PET package is 40KB. Even with gtksourceview, the combined size is one third of Geany.

USB Flash boot fail if "pkeys=de"

It has been reported that booting from a USB Flash drive with "pkeys=de" the boot failed. Without the pkeys parameter it works.

I found the problem. This is the fixed code in the 'init' script:

  vfat) #w476

   ####TODO#### need to get rid of the old 2-letter keyboard layout, so PKEYS boot

   # param works for the main kbd layouts.

   #would like to introduce codepage and iocharset, maybe as kernel params...

   #ex: codepage=850 iocharset=iso8859-2 ...no, already have 'pkeys'

   OUT_PARAM='shortname=mixed,quiet'

   [ "$MNT_O" ] && OUT_PARAM="${OUT_PARAM},${MNT_O}"

   if [ "$PKEYS" ];then

    case $PKEYS in

     de|be|br|dk|es|fi|fr|it|no|se|pt)

      modprobe nls_cp850 #v426

      OUT_PARAM="$OUT_PARAM"',codepage=850'

     ;;

	 slovene|croat|hu101|hu|cz-lat2|pl|ro_win)

	  modprobe nls_cp852 #v426

	  modprobe nls_iso8859-2 #v426

	  OUT_PARAM="$OUT_PARAM"',codepage=852,iocharset=iso8859-2'

	 ;;

    esac

   fi

   mount -t $MNT_T -o $OUT_PARAM $MNT_DEV $MNT_DIR

  ;;

...you can see where I have inserted explicit modprobe commands. Those modules should have loaded automatically, I don't know why they don't.

425: testing USB Flash booting

One or two people reported difficulty booting from USB Flash, for 4.3beta1/2.

I used the Universal Installer, installed to 1GB Kingston DataTraveler drive with vfat partition. Works. Then installed to a 2GB DataTraveler with ext2 partition. Works also.

I ran the BootFlash program and chose the default 'USB-HDD' install option. Installed to the 1GB DataTraveler. The BootFlash utility reformats the drive with a vfat partition. Works.

Puppy 4.3beta3

Close enough -- at the time of posting this, the woof tarball and the delta files are still uploading.

Puppy 4.3beta3 (425) now available, get it from here:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/test/puppy-4.3beta3/

Of course you will also be able to download from the ibiblio mirrors. MU is also hosting 4.3beta3:

http://puppyisos.org/isos/Puppylinux-official/4.3beta3/ (User: puppy Passwd: linux)

I have updated the announcement and release notes page:

http://puppylinux.com/download/release-4.3.htm

The default theme:

GTK: Polished Blue (?)
Icons: Blue Moon (ttuuxxx)
JWM: Gradient-blueish (me)
Background: Summer Clouds (author unknown)

Note, the GTK Polished Blue theme is GPL and was hosted on freshmeat.net, but it has been removed and does not seem to be available online anymore. But, we do have it as a PET package.

If anyone would like to tweak the themes, go for it. It seems reasonably ok as the theme for the final, what do you think? I don't think it's brilliant, just pleasant, though the mix of different blues may seem a bit odd.
I was thinking though, that we could do something to the Summer Clouds background -- what about integrating the Puppy logo into the cloud? No, not Superpup, although that is tempting! -- but we decided to be more serious for the final.
Anyone want to have a go at that? Here is the Summer Clouds image:

It's 141KB as it was saved at quality-factor 95, but in Puppy I saved it at 85, which reduced its size to 63KB and it still looks the same.

Forum thread for feedback on 4.3beta3:

http://murga-linux.com/puppy/viewtopic.php?t=46502

4.3final by Saturday 19th Sept.

I intend to have 4.3 final out by the end of the week.

After that, I am having a break for a couple of weeks, not really a break but doing something completely different.
I confess, I'm getting a bit burnt-out, need a change.

So guys, when beta3 becomes available, test it and I'll put in a good effort at fixing anything that needs fixing before the final.

The emphasis is now on fixing important bugs only. No upgrades unless absolutely essential.
There will of course be small bugs that can be held over for a later version.

Um yes, after 4.3 is released, we have to discuss who will coordinate 4.4. I would prefer to get back onto Woof core development and some other things. I was wondering, perhaps our old "community edition" model should be looked at again.

4.3beta3 is uploading

It's uploading to ibiblio.org right now, but they throttle transfers, so it is steady at 10.5KB/sec -- estimated completion time is 10 hours from now!

I'll announce it when done.

NicoEdit 2.0 bugs

I earlier announced that I had replaced Geany by NicoEdit 2.0, however I have to reverse that.

I found two showstopper bugs in NicoEdit 2.0:

1. No syntax highlighting for Genie code.
2. No synax highlighting at all if a file is opened from the commandline.

That second one I discovered when right-clicked on a file in Rox and chose "Open in text editor".

I am sure that Nicolas will fix them, but as 4.3beta3 is immenent I have rolled back to Geany and NicoEdit 1.6 -- the latter is a simpler version of NicoEdit that is our Leafpad-replacement.

ExpenseTracker removed

I have taken out ExpenseTracker from 4.3, as it needs work.

I took a look at it as it was reported that it crashes for any locale other than en_US.

I noticed that it displays currency as "Rs." (Rupees). I fixed that, but then I noticed that an amount I entered as "555.46" got rolled up and displayed as "555.5" -- yuck, although it does store the correct value in the database.

That could be fixed too, but I don't have the time right now.

Aqualung crashes

It was reported that if the "Enable statusbar" checkbox is unticked in the Music Store, in the Setup, then Aqualung crashes. Yes, I have confirmed that.

If someone feels like testing their latest unreleased snapshot, please do. If the bug is still there, if you feel like reporting it, please do!

Universal Installer: SCSI

I have enabled the menu choice to install to SCSI hard drive in the Puppy Universal Installer.

ALSA Wizard fixed

It was reported that the ALSA sound Wizard does not work in 4.3beta2.

Yes, this is because of the move from the full 'depmod' to the Busybox 'depmod' applet.

I have modified /usr/sbin/alsaconf so that it runs the full depmod (/sbin/depmod-FULL), with a warning displayed that it may hang on low-RAM systems.

The low-RAM problem is relieved once a swap file/partition is created, so is likely only to be a problem on the first boot, on a 64MB (or less) system.

iwlwifi firmware

Forum member lluamco reported that later firmware is needed for the iwlwifi wireless driver for the 2.6.30.5 kernel.

It is /usr/lib/firmware/iwlwifi-4965-2.ucode

I have put this into the firmware tarball, also left in the older firmware.

I have modified /etc/modules/firmware.dep.2.6.30.5, added "iwlagn.ko" onto this line:

iwlwifi:iwl3945.ko,iwl4965.ko,iwlagn.ko

PupScan fixed

Puppy 4.3beta is using Busybox 'depmod', for reasons discussed recently in this blog. However, it compromises the operation of PupScan (see System menu). Rerwin has fixed that. Extract of his forum post:

Now that the modules.*map files are no longer maintained in /lib/modules//, pupscan no longer displays the modules associated with hardware devices. I have updated it to obtain its information from the modules.alias file and added use of the dialup-modem rules file (for devices not listed in modules.alias).

I also changed it to choose that last-found module instead of the first, to match how pup_event_backend_modprobe makes its choice.

At some point, I hope to refine it so that it accounts for "pci overrides" and "preferred modules" specified in the BootManager. But this is already a little more comprehensive than when it was last working.

UPDATE 9/10/09: Turns out using the modaliases is trickier than I anticipated. I have made further refinements to pupscan (as -2), but more can be done. I may continue refinement, but please (Barry) take whatever version is here at the deadline for the next beta. Thanks,
Richard

Pstreamvid channels

In 4.3beta2, The BBC channel has audio but no video, Fox11 has neither, and NASA has audio but no video.

Probably fixable by installing Mplayer or more codecs for Gxine (or an updated xine-lib), however, I want people's first experience when using 4.3 for all the video channels to work.

Therefore I have removed BBC, Fox11 and NASA, and added SFGTV2 and SCCtv.

A great place to find out what's out there:

http://watch.squidtv.net/

Ayttm 0.5.0-146

I have upgraded to the latest from CVS.

ChangeLog:
http://ayttm.cvs.sourceforge.net/viewvc/ayttm/ayttm/ChangeLog?view=log

SeaMonkey 1.1.18

I have upgraded to the latest version.

JWM theme changing works

There was a report that running JWM Config to change theme caused /root/.jwmrc-tray to get deleted, without any backup.

I'm sorry, but it works fine for me. And there is a backup created.

If you do experience a problem, then you need to document the exact steps, and make sure it is reproducable.

Gtkdialog question

In the PPM there is a Find box with a Go button beside it. There is a request that pressing ENTER in the Find box should initiate the search.

However, if I put tags within the , they run everytime a character is typed:

   Find:

   

    ENTRY1

    /usr/local/petget/findnames.sh

    refresh:TREE1

Does anyone know how to set it so that the actions only run when the ENTER key is pressed?

e2fsprogs 1.41.9

I have upgraded this as it has important bugfixes.

As well as the shared-library package in the main filesystem, I also compiled it statically in uClibc, as the initrd needs 'e2fsck', 'fsck', and 'resize2fs'.

Intel 537 modem drivers

Rerwin has compiled these for the 2.6.30.5 kernel, and created new firmware tarball and udev rules.

I have updated, will be in 4.3beta3.

Default image viewer

I have changed the default image viewer to Viewnior, instead of MU's Pictureviewer.

Pictureviewer is still in Puppy, as MU's PuppyBrowser needs it, due to it's drag-and-drop capability.

But now, click on any image in Rox, and it will launch in Viewnior.

Note, I have stayed with Viewnior v0.6. The latest version is 0.7 which introduces image cropping, however I'm really not so sure that I want the image viewer to have editing capabilities.

EmbeddedBookmarks bugfix

MU created EmbeddedBookmarks, but when I made it into a PET package I left out a file, so that when the icon to launch EmbeddedBookmarks from within PuppyBrowser is clicked, it fails to launch. Fixed.

Gtkhash

Ttuuxxx has created a lovely little PET package for generating md5 and sha1 hashes (these are signatures that we use to verify integrity of a file).

Ttuuxx has also integrated it into ROX-Filer so that for .pet, .tar.bz2, .iso, .tar.gz and .sfs files there is an entry in the right-click menu to launch Gtkhash.

This will be in 4.3beta3.

Ttuuxxx's forum post:
http://www.murga-linux.com/puppy/viewtopic.php?t=32700

Gtkhash home:
http://gtkhash.sourceforge.net/

Hiawatha 6.17.1

I had previously discussed doing so, have now done it, upgraded the Hiawatha web server to version 6.17.1.

Hewlett Packard drivers

Thanks to MU (Mark) who created these packages. I will be placing them in the official Puppy4 repo at ibiblio.org, as it is reported to work in 4.3beta2.

These provide more complete support for HP printers and scanners than the SANE and Gutenprint drivers builtin to Puppy.

The packages will be named 'hpliplite-3.9.8-mu-p4.pet' and 'libnetsnmp-15-mu-p4' -- the latter is a required dependency.

Pfind, Pfilesearch, Pburn

Zigbert's apps upgraded:
Pfind 4.15
Pfilesearch 1.20
Pburn 3.0.4

NicoEdit 2.0

I have upgraded NicoEdit to version 2.0. Nicolas posted it here:

http://www.murga-linux.com/puppy/viewtopic.php?t=39560&start=390

This is brilliant. Nicolas, you are to be congratulated on creating such a lovely little text editor. It is good enough I reckon to satisfy most Puppians, so I have removed Geany (Geany could be in the devx perhaps).

One thing NicoEdit does not support is tabs, but there is a 'History' menu will allows quick flipping between documents -- actually, as I often have a lot of documents open, too many for the tabbed system to comfortably handle, the way NicoEdit does it is good.

Nicolas wisely rolled development back a bit, removing the theming support, and just focusing on a simple functional text editor for 2.0final.

The PET is only 47KB, although it does need gtksourceview (208KB PET) -- but gtksourceview is already needed in Puppy by Notecase.
The Geany PET package is 748KB, so we shave about 500KB off the live-CD.

GRUB fixed

Thanks to the guys reporting back in the 4.3beta2 forum thread, the problem with GRUB is fixed.

The GRUB in 4.3beta2 is patched to recognise ext4, however something is wrong with it as many people were unable to boot.

So I grabbed GRUB 0.97 from Ubuntu, release "29ubuntu53", which, if my memory serves me correctly, has a patch for ext4. I made this into a PET package and posted it to the forum. Feedback is that it works!

So, 4.3beta3 will have 'grub-29ubuntu53.pet'.

4.3 release schedule

I have got the Perth Royal Show gig coming up again this year. Those of you who have monitored my blog for over a year will know of this. I have worked there as a "Pavilion Steward" for the last couple of years, in a pavilion that has displays of the various regions of Western Australia. Here is a snapshot of my Midwest region from last year:

What I would like to do is get 4.3final out before that, and I'm aiming for the 18th September. The Royal Show starts sometime after that, but I want some leeway to handle feedback immediately after 4.3final is released -- as I won't have much Internet access while in Perth this time. I'll be in Perth about 2 weeks -- I work at the show from about 1 week before it starts, helping construction and general setup in the pavilion, then the actual Show runs for about 1 week.

So, I plan to upload 4.3beta3 about 5-6 days from now.

The alternative is that I delay 4.3final until about mid-October, but would prefer not to do that.

Perth Royal Agricultural Show home page:

http://www.perthroyalshow.com.au/

My old Cyrix CPU

I've got an old workhouse PC, mostly retired now but I fire it up every now and again to test Puppy.

Cyrix 6x86 200MHz CPU, 128MB RAM, el-cheapo 14-inch monitor. The BIOS is dated Feb. 1998.

Have just tested 4.3beta2. This is my bottom-of-the-rung machine -- if the pup runs on this, I'm happy. Yep, 4.3beta2 is fine, although I found a bug, the serial mouse setting was not remembered after creating a pupsave -- I thought that old bug was fixed, I'll take another look at it.

Anyway, tempestuous raised the question, will the 2.6.30.5 kernel be the final kernel for 4.3? I'm very happy with it, there are no showstopper bugs reported on the forum... so, yes, that's it.

The 2.6.30.5 source is here:
http://puppylinux.com/sources/kernel-2.6.30.5/

The "zdrv"

I posted this to the forum, reproducing here:

Quote from rerwin:
I am about to build a set of Intel537 drivers -- there are actually 4 of them, for the various flavors. Each is about 4 MB, so I fear for what that would do to puppy's waistline. Because I favor a puppy that can load its program squashfile into the ramfs in PCs with only 256 MB of RAM, I suspect we are facing migrating the heaviest of the modem drivers to a zdrv (or equivalent) auxiliary squashfile. Perhaps all modem drivers in their own subdirectories would be candidates for migration, except for modems built into laptops or motherboards. That would recognize that not all users need all of the modem drivers in their ramfs.

Yes, 4.3 is going to end up too fat!
I'm going to add another choice in the Woof '3builddistro' script to extract just the large modem drivers and their firmware tarballs into a separate 'zdrv'.

This zdrv will not be in the live-cd iso, it will be available separately for download.

The zdrv has to be named in a very precise manner. This is how it is done:

PREFIX1CHAR="`echo -n "$DISTRO_FILE_PREFIX" | cut -c 1`"
KERNEL3CHARS="`echo -n "$KERNELVER" | tr -d '.' | tr -d '\-' | tr -d '[a-z]' | rev | cut -c 1,2,3 | rev`"

z${PREFIX1CHAR}${DISTRO_VERSION}${KERNEL3CHARS}.sfs

In 4.3beta2 (as I intend for the final) DISTRO_FILE_PREFIX="pup"
For the 2.6.30.5 kernel, KERNEL3CHARS="305"

So, for 4.3final, the zdrv must be named:

zp430305.sfs

Anyone can provide a zdrv with this name, with their own drivers or whatever, or even start with my official one and add more stuff to it.

It isn't controlled by the BootManager, if present at /mnt/home it gets automatically loaded. Puppy will automatically run 'depmod' first time it is loaded, and again if it is removed.

FLAC, MP4 player

As requested, I have added mime-type handling for .flac audio files.
The handler for ROX-Filer is /root/Choices/MIME-types/audio_x-flac and will launch the audio file with 'defaultaudioplayer' (currently Aqualung).

Coolpup has been urging me to add mime-type handling for .mp4 video files.
Puppy already has /root/Choices/MIME-types/audio_mp4 (.m4a files) which will launch 'defaultaudioplayer'.
The MIME mechanism already recognises a .mp4 video file as video/mp4, so I have created /root/Choices/MIME-types/video_mp4 which will launch 'defaultmediaplayer' (currently Gxine).

Note that /usr/share/mime/globs already has:

audio/x-flac:*.flac
audio/mp4:*.m4a
video/mp4:*.m4v
video/mp4:*.mp4

Upgrade Flash player?

Coolpup is urging me to upgrade to the latest Flash player. Coolpup has created this information page:

http://puppylinux.org/wikka/AdobeFlashPlayer

Pup 4.3beta2 still has Flash player version 9.0.48.0, which is getting rather old.
I have stayed with it because it works for me, and is small.

So, do we really have to move up?

Are you guys finding Flash-based sites that don't work with the older player?

Of course, the latest player is supposed to have fixed some security holes.

This is the main reason I am so reluctant to upgrade:
flashplayer-9.0.48.0.pet 2545KB
flashplayer-10.0.32.18.pet 3951KB
...these are the compressed sizes, so the upgrade would cause the live-CD size to jump up about 1.4MB.

424: JWM fix, 536 modem driver

JWM version 456 PET package had an invalid tag in '.jwmrc'. Removed.

Rerwin has patched the Intel 536 analog modem (dialup) to compile on the 2.6.30.5 kernel.
I have added the source to http://puppylinux.com/sources/kernel-2.6.30.5/3rd-party/modem, also added the kernel module to Puppy, for the upcoming 4.3beta3.

Testing latest Hiawatha

I have compiled the latest Hiawatha web server, version 6.17.1.

Tested PPLOG, our personal blog, seems ok.

Very basic testing with QUISP, seems ok.

However, 4.3beta1 testers reported that they were able to get a server error. Lobster, you were one of those people.
I was wondering, do you recall the steps to reproduce that error?
If I can follow those same steps, then I will find out if the new Hiawatha has the same problem.

Advanced Xorg for 4.3beta2

MU has created an Xorg package with upgraded Intel and ATI drivers. This is reported to work in 4.3beta1 and beta2:

http://murga-linux.com/puppy/viewtopic.php?t=46266

Puppy 4.3beta2

Now available, download from here:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/test/puppy-4.3beta2/

Theme
Still not finalised, though the choice in 4.3beta2 looks ok to my eyes. Very business-like. Using icons by steve_s and background by zigbert.

Kernel
I have built two live-CD iso files only, both with the 2.6.30.5 kernel. One of them has the SCSI drivers.
I am seriously thinking of offering 4.3final with only this kernel -- other people can create builds with older kernels if they wish, and offer them for others to download.
4.3beta2 with 2.6.30.5 has heaps of analog (dialup) modem drivers provided by rerwin (including all the Conexant drivers), which is why it is a bit fat (101MB).

Note, it will please a certain person, this kernel also has the Ham Radio modules.

Image viewer
Surprise -- Fotoxx is gone, replaced by Viewnior. Let me know what you think. Do you think that Viewnior should be the default image viewer instead of PictureViewer?

Text editor
Back with Geany. We still have syntax highlighting for Genie code via the NicoEdit 0.16 basic text editor included in the 4.3beta2 build.

Printing
Would someone like to volunteer to create a hpijs printer package? The reason I ask this is there has been feedback that existing hpijs pet packages do not work in 4.3beta1. Some of you guys who are much more into messing round with printing could do a better job of preparing a hpijs pet package than me.

Release notes
The release notes/announcement page is still there, but I haven't updated it since 4.3beta1:

http://puppylinux.com/download/release-4.3.htm

Known bugs in beta2
1. Nathan Wallpaper Setter broken. The XML for gtkdialog is broken by the absence of fotoxx. The fixed package is 'wallpaper-0.5.4.pet', available from the forum 4.3beta2 feedback thread.
3. Even as I was uploading beta2, zigbert brought out another Pburn (3.0.4) -- grab the latest from here: http://murga-linux.com/puppy/viewtopic.php?t=23881

Forum feedback
For all bug reports, and good news too:

http://murga-linux.com/puppy/viewtopic.php?t=46273

Planning for 4.3beta3
I intend that this will be the last beta before the final. Some things that need to be done by beta3:

1. Fix upgrading a 'pupsave' from 4.2x.
2. Rerwin's 536/537 modem drivers.
...more...

GenieCC Genie code generator

I'm posting this here, as the main forum is down.

Nicholas has created a "code generator" for Genie. What it does is translate a simplified pseudo-code into Genie code.

If the forum was working, you could download it. Maybe it will come-good soon.

Anyway, to compile it in 4.3beta1 or beta2 requires a couple of modifications:

Build it:

# ./waf configure
# ./waf build

The configure step reports that Vala compiler 0.7.5 or later is required. Puppy only has 0.7.4. However I edited 'wscript' so that it would accept 0.7.4, and it compiles.

However at the second step it was unable to find the gee header files. They are at /usr/include/gee-1.0/gee. So, I created a link:

# ln -s gee-1.0/gee /usr/include/gee

The idea is interesting. If this pseudo-code is simple yet capable, and well documented, then it would make it easy to get into creating Genie apps.

Murga forum problems

John has contacted me about the main Puppy forum. It is having trouble. John is putting in considerable effort to try and fix it -- if not, may move to a new host in a couple of weeks.

John is one of those guys who supports Puppy behind the scenes. It is no small feat to look after such a big and busy forum. John also hosts it at his own expense.

We do of course have the linuxquestions.com alternative forum for Puppy. I think Lobster create one also?

Forum member tclhost (Thom) has just contacted me. He has setup another backup forum:

http://puppylinux.us/smf/

...I emailed and asked if some more categories could be created in it.

Our backup LinuxQuestions forum is here:

http://www.linuxquestions.org/questions/puppy-71/

4.3beta2 in 10 hours

I had it ready several hours ago, but was uploading the latest PET packages to ibiblio.org.

Now I have started to upload the 4.3beta2 files, but you'll have to wait.... and wait. Ibiblio can be painfully slow sometimes -- currently uploading at 10KB/sec and estimated completion is 10 hours from now...

I'm posting this message at 7.35am GMT+8 (Western Australia). So, barring any upload failure, completion will be by 5.35pm.

Use a international time converter, such as this one, to figure out your local time:

http://www.timeanddate.com/worldclock/converter.html

...for example, in Los Angeles the completion time will be "Sunday, September 6, 2009 at 2:35:00 AM"

Of course, I'll announce it on this blog.

ALSA upgrade, scanner, printer tests

Audio
As has been reported, alsa-lib and alsa-util in 4.3beta1 are rather old (v1.0.16). I have upgraded them to v1.0.20, matching the version in the 2.6.30.5 kernel.

Immediately there is a good outcome. One problem has been that, with certain audio hardware anyway, small audio files get truncated. For example, /usr/share/audio/2barks.au and 2barks.wav come out as a single bark. No more, on my laptop at least, I now get two distinct barks.

Scan
Recently when I was driving home from Perth, I stopped at my favourite little opportunity shop -- I have mentioned this place before, picked up a couple of PCs there before. It's in a small country town, run by the local church.

Anyway, they had a USB scanner, Hewlett Packard ScanJet 4200C, for $4. That's a bit of a gap in my test suite, so I bought it.

This evening I am doing some testing of the soon-to-be-released 4.3beta2 and tested this scanner. I ran Xsane, and scanned a page, no problem whatsoever. Very nice.

Print
Have also done a basic sanity test for printing, to my parallel-port Epson Stylus C41SX, that was also acquired from the opportunity shop awhile back.

As I recall, there was feedback that printing was not working from certain applications, and over a LAN. I will have to hold-over the latter as I don't have a LAN setup.

First stop, trying ePDFView. There's a demo PDF file at /usr/share/examples/ps-pdf/Acrobat.pdf. Yep, prints.

Now for Abiword. This is version 2.6.3 with 2.6.8 plugins, as discussed in my earlier post. I opened /usr/share/examples/text/testdoc.doc. Success again.

And SeaMonkey. Opening /usr/share/doc/home.htm. That prints also.

After I have released 4.3beta2, would someone like to volunteer to create a hpijs printer package? The reason I ask this is there has been feedback that existing hpijs pet packages do not work in 4.3beta1. Some of you guys who are much more into messing round with printing could do a better job of preparing a hpijs pet package than me.

423: Abiword 2.6.3

I have been experimenting with Abiword 2.6.3 and the 2.6.8 plugins. Seems to be working fine.

One interesting discovery, that would explain why the Presentation plugin did not work in earlier testing with Abiword 2.7.x -- it requires the Loadbinding plugin.

I compiled Link-grammar 4.6.1, as I would like the grammar plugin to be working.

Here is how I compiled everything:

abiword 2.6.3

error, chooses internal goffice, so edit 'configure', change "libgoffice-0.4" to "libgoffice-0.6":



GOFFICE_VERSION_REQUIRED='0.4.0'

goffice_modules="libgoffice-0.6 >= $GOFFICE_VERSION_REQUIRED"



# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --enable-printing --enable-threads --disable-gucharmap --disable-scripting --without-ImageMagick



link-grammar 4.6.1

# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --build=i486-pc-linux-gnu



abiword-plugins 2.6.8

# ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --build=i486-pc-linux-gnu --disable-all --enable-abigrammar --enable-freetranslation --enable-gdict --enable-presentation --enable-bmp --enable-jpeg --enable-librsvg --enable-mswrite --enable-OpenDocument --enable-OpenWriter --enable-pdf --enable-xhtml --with-abiword=../abiword-2.6.3 --without-boost --enable-loadbindings

Note, I didn't compile the OpenXML plugin as it needs the 'boost' package. Note, technosaurus has compiled Abiword 2.7.10 with OpenXML, and found that it only needs one small library out of boost. I suppose we could add the OpenXML plugin to my package afterwards.

Fatdog2 32bit beta2

This was released quietly on 8th August, I only noticed it a few days ago. Kirk has used T2 to compile from source, and the Woof build system, to create Fatdog2, in two flavours, 32-bit and 64-bit.

This is extremely interesting work, and likely to be one of the Puppy-5 releases.

http://murga-linux.com/puppy/viewtopic.php?t=45335

KolibriOS: a tiny operating system

Wow, this has stirred up some memories! See the feature story in this week's Distrowatch News:

http://distrowatch.com/weekly.php?issue=20090831

Here is the KolibriOS home:

http://wiki.kolibrios.org/

KolibriOS is written in x86 32-bit assembly language and is a fork of MenuetOS. Back in 2002 I was involved with MenuetOS, and still have a web page:

http://goosee.com/menuetos/index.html

What about the original MenuetOS project? Yes, it is still going:

http://www.menuetos.net/

Fascinating! It's a great little OS, of course has extreme hardware compatibility limitations, and is reinventing the wheel, but it's design is so satisfying from the point of view of the tiny size and extreme speed and efficiency.

e3 fixed

I was puzzled, when I released a new 4.3 alpha or beta, e3 was broken. Note, 'vi' runs e3 in its vi-compatible mode.

I have found out why. I introduced file stripping into the '3builddistro' script in Woof, and it also strips e3 -- result, e3 reduces from 13KB to 208 bytes and won't run.

aufs2 utilities

I downloaded the Aufs2 utilities from GIT today, and created pet package 'aufs2-util-20090904.pet'.

These utilities will be in 4.3beta2. This includes /usr/share/doc/aufs.htm.

Puppy has been using Unionfs for sometime, previously when we used Aufs it was the 1.x series.

We will probably need to study Aufs2 to get the maximum benefit from it, but that can probably be a target for pup 4.4:

Clean shutdown
For example, it looks like there is potential for a clean shutdown, something that is an ongoing issue for Puppy.

Write directly to flash
The /usr/local/petget/installpkg.sh script is able to write directly to the save-layer when running Puppy on a Flash drive, in the case of Unionfs.
We never had this working properly with Aufs.

Unipup
My Unipup concept, that I was working on sometime ago, runs Puppy totally in the initramfs and does not perform a switch_root. It seems that Aufs2 may be able to support building a layered filesystem within the initramfs -- this capability was a configure option when I compiled the kernel module, which I enabled.

EDIT, 6 SEPT 2009:
Dynamic load
Some of the guys have been working on adding and removing SFS files without rebooting. They had this working with aufs1.

PPM 'ALL' category

By popular demand, I have reinstated the 'ALL' category in the Puppy Package Manager main GUI window.

3G wireless, agrsm modem

Rerwin has done a lot of work helping people with getting their 3G wireless devices working. He posted an upgrade to the 4.3beta1 forum thread. These files are upgraded:

/sbin/pup_event_backend_modprobe
/usr/bin/modemprobe, modemprobe_erase, modemtest, pupdial, usb_modeswitch.sh

Rerwin has also compiled the agrsm modem driver for the 2.6.30.5 kernel. He has patched the source to get it to compile, which I have uploaded here:

http://puppylinux.com/sources/kernel-2.6.30.5/3rd-party/modem/agrsm-20090502-patched.tar.gz

There is a new firmware tarball:
/lib/modules/all-firmware/agrsm.tar.gz

Kernel modules:
/lib/modules/2.6.30.5/extra/agrmodem.ko, agrserial.ko

Udev rules have been upgraded:
/etc/udev/rules.d/60-udev-modem.rules

SFS Converter 1.3.2

Trio has written a little GUI application that will convert a SFS file from version 3 to version 4 and vice versa. This is to be found in the Utility menu.

I made a few small mods and bumped the version number slightly. The PET package is now 'sfs-converter-1.3.2.pet'.

When you click on a SFS file in Rox, it will be mounted (via /usr/sbin/filemnt), however if the wrong version for the current kernel an error message will come up. I have not put in automatic launching of the converter program, instead just added to the error message that a SFS-version-converter is to be found in the Utility menu.

The BootManager (/usr/sbin/bootmanager) puts up an information box listing any SFS files found at /mnt/home that are the wrong version. I have also added an information message that a SFS-version-converter is to be found in the Utility menu.

Sysprof 1.0.12

Sysprof is a "system profiler", very useful for analysing resource usage of an application. I have upgraded from 1.0.10.

Sysprof consists of a kernel module and a GTK GUI excutable. The application will be in the 'devx' SFS file.

423: Packages upgraded

Trio's JWM Theme Maker to 1.4
Zigbert's Pburn CD/DVD burner to 3.0.3
Zigbert's Pfind file finder to 4.14
Tasmod's Psync time server synchronizer to 1.0
Trio's You2pup youtube downloader to 1.3-1

4.3beta2 schedule, updated

Some major bugs have been fixed, and things are looking pretty good.

Just to let you know, I'm currently thinking of uploading 4.3beta2 in a couple of days. Let's say, Saturday or Sunday.

423: GRUB fixed

4.3beta1 testers have reported that a frugal installation of Puppy failed. There was a problem with installing GRUB.

If you already had GRUB installed and all that you needed to do was modify the /boot/grub/menu.lst file, then you would not have encountered this bug.

On the otherhand, if you ran /usr/sbin/grubconfig, either from the Universal Installer or the System menu, to install GRUB, then you would have hit the bug.

There was a typo at line 1003.

I fixed it, but then I discovered an awful bug that was in the original script -- it wasn't written by me, I'm just hacking on it!
The partition to which GRUB is to be installed gets mounted on /tmp/boot, however when grubconfig failed it also left the partition mounted. After fixing the script (line 1003) and rerunning the script, one of the first things it does is wipe everything in /tmp/boot -- meaning, it wiped my partition!

There are a couple of places in the script where this can happen. I have put in some safety checking, so this potential disaster is now fixed.

Fortunately I didn't lose a vital partition!

423: libshadowfb.so

Forum member playdayz made this request:

would it be possible to include libshadowfb.so with the basic Puppy? It is included in xorg_xorg_full (for 4.2.1); it is necessary for the two virtualizers vmware and virtualbox to use their own video drivers. If it is present in /usr/X11R7/lib/xorg/modules then vmware works out of the box with its own vmware_drv (which is included in the basic Puppy). If it is not present then X will not start from xorgwizard because xorgwizard chooses the vmware_drv which will not start without libshadowfb.so. thanks

Done.

2.6.30.5 packages

I have uploaded the 2.6.30.5 patched source and 3rd-party modules here:

http://puppylinux.com/sources/kernel-2.6.30.5/

The PET package that is used in Woof to build a puppy is here:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/pet_packages-woof/linux_kernel-2.6.30.5-tickless_smp-p4.pet

There are some more 3rd-party modules that could be compiled. I was unable to compile these:

acx wireless driver
linux-wlan-ng wireless, USB module only
536/537 modem
ess modem

Rerwin said that he would investigate compiling the Intel 536/537 drivers.
Regarding linux-wlan-ng, the hostap drivers already in the kernel supercede that, but I think the USB module in linux-wlan-ng is still required?

Conexant modem drivers

Rerwin (Richard) has done a lot of work in getting the Conexant drivers to compile and making them auto-configure in Puppy. You can find his posts in the forum 4.3beta1 feedback thread.

I have compiled the Conexant 'dgc', 'hcf' and 'hsf' drivers for the 2.6.30.5 kernel, and updated rerwin's firmware tarballs accordingly.

I'm uploading files right now.

2.6.30.5 kernel!

I decided to test the 2.6.30.5 kernel, see if the multisession problem using cdrkit or cdrtools is fixed. Yes it is!

I have tested a CD-R and a DVD-R, multisession works fine.

As I have reached 4.3beta1, I don't want to make any fundamental changes. But, this is extremely tempting...

I patched the 2.6.30.5 kernel with aufs2 (cvs 20090831) and it no longer supports unionfs compatibility-mode. Meaning, the whiteout files have different names. This affects a lot of scripts, such as snapmergepuppy, savesession-dvd, init, rc.update. I'm reluctant to make fundamental underlying changes at this stage, but I'll check it out, might not be potentially troublesome.

I'll examine the situation for the whiteout files this afternoon. Also I'll compile some more 3rd party drivers.

Note, I only did three patches to the kernel. The loglevel patch, aufs2, and an ALSA patch provided by rerwin.

b43 firmware problem solved

I think so anyway. I would like to thank trio for patiently answering a series of questions. I would have loved to have had the actual hardware to test, but I reckon that I have got to a solution.

I will need to study the evolution of the uevent/udev mechanism to fully understand this. I have been relying upon the existence of 'modalias' files in /sys for replaying kernel uevents. In /etc/rc.d/rc.sysinit there is code that locates the modalias files, then looks for a 'uevent' file at the same location then sends a "replay" message to it.

The problem is that it doesn't always work. I don't know if the problem is due to an evolution in the mechanism in recent kernels, or not. I'm currently running 2.6.29.6, I would need to check this on older kernels.

For USB interfaces, see this:

# cat /sys/bus/usb/devices/*/uevent | grep 'MODALIAS'

MODALIAS=usb:v1D6Bp0002d0206dc09dsc00dp00ic09isc00ip00

MODALIAS=usb:v1058p1010d0105dc00dsc00dp00ic08isc06ip50

MODALIAS=usb:v1D6Bp0001d0206dc09dsc00dp00ic09isc00ip00

MODALIAS=usb:v1D6Bp0001d0206dc09dsc00dp00ic09isc00ip00

MODALIAS=usb:v046DpC03Dd2000dc00dsc00dp00ic03isc01ip02

MODALIAS=usb:v1D6Bp0001d0206dc09dsc00dp00ic09isc00ip00

MODALIAS=usb:v1D6Bp0001d0206dc09dsc00dp00ic09isc00ip00

# 

# cat /sys/bus/usb/devices/*/modalias                

usb:v1D6Bp0002d0206dc09dsc00dp00ic09isc00ip00

usb:v1058p1010d0105dc00dsc00dp00ic08isc06ip50

usb:v1D6Bp0001d0206dc09dsc00dp00ic09isc00ip00

usb:v1D6Bp0001d0206dc09dsc00dp00ic09isc00ip00

usb:v046DpC03Dd2000dc00dsc00dp00ic03isc01ip02

usb:v1D6Bp0001d0206dc09dsc00dp00ic09isc00ip00

usb:v1D6Bp0001d0206dc09dsc00dp00ic09isc00ip00

That is,the MODALIAS information is to be found in both the 'modalias' file and the 'uevent' file.

The problem arises with some newer interfaces, there is no 'modalias' file, only a 'uevent' file, and the MODALIAS variable is only to be found in the 'uevent' file.

This is the situation that trio has reported:

There is no modalias anywhere. But found two uevent(s)

/sys/bus/ssb/devices/ssb0:1/uevent and /sys/bus/ssb/devices/ssb0:0/uevent

Inside the uevent file:
MODALIAS=ssb:v4243id0812rev0A

I have modified /etc/rc.d/rc.sysinit accordingly. It should now successfully replay the uevent for the network hardware that requires the b43 module, and the firmware should get installed.

Video test collection

I have uploaded some of the video files that I use for testing media players in Puppy:

http://puppylinux.com/test/media-test-video/

If you have compiled a new release of ffmpeg, xine-lib, Gxine, Mplayer, or whatever, and want to assess it's suitability for use in Puppy, this collection is useful. All of the above files play with both audio and video in Puppy 4.1.2 and 4.3beta, which has a fairly old version of ffmpeg and xine-lib.

My testing with more recent releases of ffmpeg has given patchy results. One thing to watch out for is choppy audio playback (to the extent that it is unintelligible) -- especially with the Lindows RealMedia file. Some media players may also have missing audio or video with certain files -- even if you have all the required codecs.

Video DVDs should also be tested. In particular, the media player needs to be able to display and navigate the initial menu that most DVDs have -- Mplayer is an example that fails dismally here, all versions.

This file has links to further audio/video test files:

http://puppylinux.com/test/media-test-video/sites.txt

P.S.
IT_Pro.wmv is very amusing!

"fontconfig development files missing"

Coolpup recently informed me that the development files for fontconfig were missing in 4.3 pre-beta. I replied that they are present.

Right now I'm compiling WINE, and when I ran the configure script it complained that fontconfig and libGL development files are missing. This reminded me of the earlier report from coolpup. I double-checked, they are present.

The problem is that the configure scripts of some applications are extremely dumb. Even though the '/usr/lib/pkgconfig/fontconfig.pc' informs that the development files are to be found under /usr/X11R7, the configure script won't look there.
However, if you run these export operations before running configure, that should fix it:

# export CPPFLAGS='-I/usr/X11R7/include'
# export LDFLAGS='-L/usr/X11R7/lib'

There aren't many distros that use /usr/X11R7. The Puppy 4.x series does because the original packages were compiled in T2 and that's how T2 does it. Ditto for kirk's Tpup.

Puppy with WINE builtin

A couple of days ago I gave a Puppy 4.21 CD to a local farmer. Yesterday he came back, quite excited. He said that he booted the CD on his PC and his son's PC and everything worked, and he got onto the Internet straightaway no hassles.

The reason that he came back to see me, is he wants to run Windows applications in Puppy. That's not something I'm into, but I suggested the easiest route for him might be to download a Puppy derivative with WINE builtin.

The only one I know of is tazoc's Lighthouse pup:

http://murga-linux.com/puppy/viewtopic.php?t=38898

...which looks real nice from the description and snapshots.

Does anyone know of any other puplets with WINE builtin, or alternatively a pre-prepared WINE pet that will just install and run without requiring any setup?

Anyway, I'll send an email to the farmer, let him know about Lighthouse pup.

Planning for 4.3beta2

Thinking ahead, for the benefit of those who are eagerly awaiting the next release, I can probably upload 4.3beta2 early next week, say Monday or Tuesday.

The Unionfs problem is still unresolved. Looking back in my blog, July 2008, I reported a problem with Unionfs version 2.3.3 which forced me to go over to Aufs:
http://www.puppylinux.com/blog/?viewDetailed=00207
The developers were working in version 2.4, a major change, but for a few months they totally ignored bug reports for the 2.3 series. I was most unhappy about that.
Regarding the latest problem, I submitted a bug report on Aug 22, so hopefully it will get attention soon:
https://bugzilla.fsl.cs.sunysb.edu/show_bug.cgi?id=634

I think by this coming Sunday I will have to recompile the 2.6.29.6 kernel, I am very much hoping for the last time, and if Unionfs is not fixed by then I'll leave it out. Regardless, I'll put in the Aufs patch.

Regarding the 'devx' file, the gtkdialog examples were missing in 4.3beta1, fixed. Also, the next release will have the 'git' package.