logoPuppy developer news:

4.00alpha3 to 4.00alpha5
left-arrow Older news

 arrow-rightLater news

Puppy Dingo alpha5 uploaded

January 11th, 2008

Get it from here:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/test/puppy-4.00-dingo-alpha5/

If I haven’t announced a bugfix for something in this blog since the earlier alphas, then it isn’t fixed yet. I always try to report here when I fix something.

I’m not sure that multisession works properly (beyond saving one session), as I am not yet home and able to test on my own PC.

I uploaded alpha5 just before seeing MU’s post about synchronising net-setup.sh with rerwin. So, if you want to do anything with that script, please get the latest from here:

http://www.murga-linux.com/puppy/viewtopic.php?t=25098


8 Responses to “Puppy Dingo alpha5 uploaded”

  1. Sage Says:
    January 11th, 2008 at 11:47 pm

    There is good news and bad news!
    IDE HD is now visible, the PUI works and the GRUB installer works (although the curious tray opening is back with a new feature - it stays open!).
    And now for the bad news.
    The boot up looping is back. Sorry - I can’t remember the answer and probably wouldn’t understand it either.

    Cfdisk is still dead in the water.

  2. Sage Says:
    January 12th, 2008 at 12:50 am

    It gets worse. The wrong menu.lst is written - it alludes to fd0 after the opening lines when only the HD should be listed. Although there seems to be confusion between hda & sda, editing these does not stop the looping - the problem may not be in menu.lst?

  3. pdrito Says:
    January 12th, 2008 at 9:26 am

    Barry,

    I confirm, multisession CD / DVD now is working on Puppy Dingo alpha 5.
    I load the iso first time, no problem.
    I make internet setup, saving on my DVD +RW , all went OK.
    I reboot and all setings and saved appears in the second boot.
    I make additional setings , adding firefox, transmission, linuxrar, and others. Saving again on DVD +RW.
    I reboot and puppy Dingo alpha 5 load the 2 Directories saved and all went OK.

    Bravo. We will have puppy forever.

  4. Sage Says:
    January 12th, 2008 at 2:00 pm

    Sadly, there is more - can’t ’see’ other new installed distros which use kernel 2.6.23 on IDE with Dingo, but no problems with v3.01.

  5. BarryK Says:
    January 12th, 2008 at 6:02 pm

    JustGreg has started a forum thread for feedback on alpha5:
    http://www.murga-linux.com/puppy/viewtopic.php?t=25348

  6. Sage Says:
    January 12th, 2008 at 7:43 pm

    Barry: If you prefer to receive all comments via John’s Forum when you announce important changes/revisions/etc it might be advantageous for you yourself to start a new thread? Being under your own name, everyone would then know that it was kosher and would receive your attention.

  7. nic2109 Says:
    January 12th, 2008 at 9:03 pm

    Sage; that’s a really good idea.

    I for one have been confused as to just where to post what.

    To complete “the set” as it were, a thread for suggestions/comments specific to the release in addition to one dedicated to bugs might be handy.

  8. Jonnijon Says:
    January 13th, 2008 at 7:28 pm

    Barry, Thank you generally for your dedication and efforts at Making Puppy all that it can be. Also, Ditto as to agreeing with Sage and nic2109 as to just where to post suggestions/comments for both confirmation of success or as to bugs new or confirmed. I for one have remained silent for many months because of this confusion!

mhWaveEdit, NoteCase updated

January 11th, 2008

Recent Dingo alphas have mhWaveEdit version 1.4.13 and there is feedback that it has bugs. So, I have upgraded to the latest, v. 1.4.14.

I have upgraded NoteCase to version 1.7.6.

Forum member ‘cb88′ has recommended that the ‘alsamixer’ be included in the menu, not just ‘zmixer’, as it has far more controls. Okay, done.

Forum member ‘rerwin’ has made another fix for ‘rc.network’ (see recent blog post for the original news on this) which works in conjunction with a fix to the net-setup.sh script.

Note, I haven’t yet applied MU’s localisation enhancement for net-setup.sh, as we first have to synchronise with the changes that rerwin has made. That is, MU worked on the script version before rerwin’s recent changes to it.


2 Responses to “mhWaveEdit, NoteCase updated”

  1. MarkUlrich Says:
    January 11th, 2008 at 9:59 pm

    Barry, the last message in that thread now contains net_wiz_clientid_fix+localization2.tar.gz
    This merges rerwins addons and my localization.
    So we are syncronized now :-)

    Mark

  2. MarkUlrich Says:
    January 11th, 2008 at 10:42 pm

    Mark,
    Thanks for responding to Barry’s blog re net-setup. Would you also mention there that we have not been able to include the 3.95 fix, because we cannot find it?

    I have not commented yet in his new blog, so am not authorized to comment.
    Thanks.
    Richard

Fotox upgraded to v. 34

January 11th, 2008

Fotox is v. 27 in Dingo, I have upgraded it to v. 34.

I’ll probably upload Dingo alpha5 within 24 hours of now. I have to upgrade a few more things, and I have just discovered a peculiar bug when booting from USB — which must be fixed before the alpha5 is released. I have to go and do some handyman stuff at my relative’s place right now, so will get back to this in the afternoon.

The first spam that got through

January 10th, 2008

Today there was a comment that was obviously spam. It was not from a registered user.

I have two layers of protection in this blog. First, users must register and get a password via email. This is a piss-weak protection as all the spammers can bypass it and post comments without registering.

So, I implemented a second layer of protection, that the first comment a user posts must be confirmed by me. This has worked, although it does mean I have to manually delete spam posts. But, they are all in the queue waiting to be confirmed, so it’s easy to tick the boxes to delete them.

Until today. I got the first spam that has bypassed both of these defenses. I only saw it when looking at the “Recent comments” sidebar.

Probably time to upgrade WordPress, introduce a graphic captcha system. Or whatever. Perhaps I’ll have to turn off commenting while I’m on holiday.


3 Responses to “The first spam that got through”

  1. Dougal Says:
    January 10th, 2008 at 8:27 pm

    I recall seeing on some Wordpress-based blog that they had at the bottom something like “using Dave’s spamcatcher for Wordpress” or something… it’s some kind addon for Wordpress.

  2. raffy Says:
    January 10th, 2008 at 11:01 pm

    Funny, I always see “You must login to post a comment”. I hope you have activated the Akismet plugin (it’s in the Plugins menu). It requires registration, but it is quick and free.

  3. RobertB Says:
    January 11th, 2008 at 6:34 am

    I don’t think there’s any defense against a truly determined spammer, just like there’s not much of a defense against someone who wants to break into your house. It’s just a matter of making your house less appealing than the one next door. I think you’ve done that, by requiring registration and putting the first comment in a moderation queue — that’s what I’ve done with the email discussion lists I own.

    But every once in a while, someone will act like they’re part of the community long enough to get off of “moderated” status, and they’ll post something stupid. There’s just not much you can do but delete them after the fact. Once every month or so is still better than wide-open, though.

Multisession now working …now broken

January 9th, 2008

In Dingo alpha4 “full” build (Jan 5), I changed from cdrkit to cdrtools (I don’t recall if I had already made the changeover in barebones alpha4). Anyway, there was a typo which caused the cdrtools package to be left out of the build. Result, neither cdrkit or cdrtools is in the build. No wonder multisession doesn’t work. Burning of any kind won’t work.

So, I put in cdrtools, tested multisession, and yes, it did save the session, but did so incorrectly. The command is something like this, simplified here:

growisofs -M /dev/sr0 -D -R -quiet -graft-points 2008-1-8-11-22=/initrd/pup_rw

This is supposed to create a folder on the DVD named ‘2008-1-8-11-22′ with the saved session in it, but instead the session got saved at ‘/’.

Well, growisofs is a frontend for mkisofs, the latter being a utility in cdrtools. So, that’s where the buck stops. Those who have followed my blog for sometime may recall earlier grumbles about mkisofs. Puppy 3.01 has cdrtools version 2.01.01a23 but the mkisofs utility did not work at all, just exited with “unknown error”. So I made up a hybrid package for 3.01, named ‘cdrtools-2.01.01a23-mkisofs_01a10, which has mkisofs from cdrtools 2.01.01a10.

Now at cdrtools version 2.01.01a36, we have a mkisofs that does something, no error message, the only problem being that it does the wrong thing.

So, I went back to cdrkit (v1.1.6), the Debian fork of cdrtools, and this time the session saved nicely in a subdirectory on the DVD just like it is supposed to. Rebooted, the saved session reloaded just like it is supposed to.

Then I saved a second session, rebooted, but this time got the “cannot find pup_395.sfs” message and got dropped to a shell in the initramsf. I tried ‘mount -t iso9660 /dev/sr0 /mnt/data’ and got a “failed” message.

I took the DVD to my laptop and mounted it with Pmount, no problem, and there are the two saved sessions, as expected.

So, this is a “progress” report. At this point in time I’m mystified why the DVD won’t mount while in the initramfs. I need to rest now, will continue trying to solve this mystery later.


3 Responses to “Multisession now working …now broken”

  1. pdrito Says:
    January 9th, 2008 at 9:53 pm

    nice Barry !!

    It is a good news that multisession are in progress now.

    By the way, I want to put here something that, I think, is relationed to multisession issue.

    As I am a NO HDD PC puppy user, I only depend on CD/DVD multisession work.
    The fact is that till Puppy 2.14 and Puppy 2.15CE version, all the saved session put pup_214.sfs and zdrv_214.sfs on my working puppy “/ directory”.
    So, if I want to add software like new_2.14.sfs to puppy, I put in the same “/ directory” and make a save on CD/DVD reboot.

    Unfortunately, from puppy 2.16; puppy 2.17.1; puppy 3.0.1 and puppy 4 versions, this characteristic of puppy is NO longer available.

    As I depend strongly on CD/DVD multisession puppy, I beg you to look into this issue.

    Thank God , we have puppy forever.

    pdrito.

  2. BarryK Says:
    January 10th, 2008 at 5:34 pm

    Sadly, the problem of being unable to mount a multisession DVD appears to be related to libata. It also depends on the hardware.

    Right now I’m away from home and only have two computers to test on, one owned by my relative, the other my Acer laptop. I have never been able to boot a multisession DVD on my laptop, so can only test multisession on one PC, and as reported that failed to mount the DVD at bootup after creating a second saved session.

    If I boot a normal Puppy live-CD on my relative’s PC, then insert the multisession DVD, it will not mount, with error messages in /var/log/messages.

    However, the same multisession DVD mounts in my laptop, no errors in /var/log/messages.

    I googled around and found others getting these error messages. There was one interesting Ubuntu forum thread (mid to late 2007), where people had upgraded then were no longer able to access their data DVDs. The only solution these people came up with was to go back to an older “pre-pata-libata” kernel.

    However, the problem seems to be restricted to only some libata PATA hardware drivers. I also booted our earlier Dingo alpha that has the 2.6.24-rc4 kernel, but same problem.

  3. BarryK Says:
    January 11th, 2008 at 7:43 am

    Ha ha ha!!! I should have tried this test earlier, but sometimes the obvious isn’t so obvious at the time.

    I booted Puppy 3.01 with pfix=ram on my relative’s PC, which of course has the conventional /dev/hd* notation for IDE drives. Then I tried to mount my multisession DVD (with two sessions saved on it) — and it wouldn’t mount, same errors in /var/log/messages.

    So, PATA libata is not to blame. The problem is more generic than that.

    This means that we are back in business with multisession. When I get home I’ll test on my usual PC in which I test multisession, that can mount multisession DVDs.

    Hmm, I wonder what is to blame, the IDE driver or the make/model of DVD drive itself?

Testing frugal install

January 8th, 2008

With the latest build of Dingo on CD, I booted a PC, chose ‘pfix=ram’, then ran the Universal Installer and did a frugal install. It went without any hitch. I installed into a ext3 partition, in directory ‘puppy395′. The Uni Installer recommended an entry for GRUB, that I put into the ‘menu.lst’ file:

title Puppy Linux 395 frugal
rootnoverify (hd0,2)
kernel /puppy395/vmlinuz pmedia=atahd psubdir=puppy395
initrd /puppy395/initrd.gz

I removed the CD and rebooted, the frugal install came up fine. At shudown I created puppy395/pup_save.2fs and at bootup it was recognised and used. No problems whatsoever.

Somebody mentioned that they did not get the offer at shutdown to save to the entire partition — I did, though I chose the pup_save file — but of course the partition is ext3 so this “full partition” option is offered.

I was expecting trouble, from the bug reports in the various alphas, and was surprised that it just worked. Notice above, ‘pmedia=atahd’ — that is because of the new unified IDE/SATA, but if ‘idehd’ or ’satahd’ had been specified on the boot commandline they would still work. I haven’t done anything to fix any bug with frugal install, so if anyone who did experience a problem is reading this, did you do anything different from the above, or have some different situation?

Regarding sage’s report on invisible IDE hard drives, I haven’t got this problem on the five PCs that I usually test on. These range from an ancient PC with Cyrix K6×86 200MHz CPU upwards. In fact, I just tested the old Cyrix today, to checkout the Zip drive — works fine, driver support for IDE Zip is now built-in to the kernel, no module to load.

Regarding inivisible USB drives at bootup, well that’s fixed, see recent blog post.

Tomorrow I’ll test on a couple more PCs.


3 Responses to “Testing frugal install”

  1. charnisingh Says:
    January 8th, 2008 at 8:19 pm

    when i first started with puppy it was puppy 2.10, grub for dos was installed, since then i have used all the puppy versions as frugal istall on fat partition using only 4 files specified by you, save file was made automacly, since then i do not have any problem even many a times the save file was upgraded . the present file i am using is aupgrade from farfree pup to dingo barebone 394 but no problem at all.
    thanks for such a simple way of using puppy.

  2. Sage Says:
    January 9th, 2008 at 2:35 am

    The mystery deepens, then!
    It has warmed up a little in GB this last couple of days so I might be able to open up my attic room where I still keep an old K6-II/500 and a Cyrix-400 (had those beasts sent over from Uncle Sam - never released in the UK). Got an old PII slot1/ C400 somewhere, too. Most of my early testing is on late-ish SktA boards and a couple of P4/2.2-2.4 reserved for checking out anomalies (both failed the Dingo4alpha4 test).
    Unfortunately, however, there is little doubt about my results because my immediate reaction was to run the discs up with P3.01/2.17/1.0.8r1 - none of which created invisibility!

  3. Sage Says:
    January 9th, 2008 at 1:10 pm

    Making some progress with diagnosis of the absent IDE s and will report later in detail on the Forum. But make no mistake, Dingo is badly broken wrt to IDE in a way that doesn’t occur with any previous release. For folks like y.t. who depend on Full installation, this is terminal failure.

Xorg Wizard rerun-at-every-boot bugfix

January 8th, 2008

This one has been around for a long time. There have been reports on the forum, people complaining that they have to rerun the Xorg Wizard at every boot, even though they have a ‘pup_save’ file or have a full HD install.

I found the cause of the problem. Puppy uses hardware profiling for video, so that if Puppy is booted on a different PC or a different monitor, the Xorg Wizard is rerun. A ‘xorg.conf.xxx’ file is created in /etc/X11 for each hardware profile, where ‘xxx’ is a string that represents the video chip and the monitor.
Puppy runs ‘ddcprobe’ to return information about the video chip and the monitor, but it is the latter that is the cause of the bug.

The problem is the ‘eisa:’ parameter returned by ddcprobe, that is used for the monitor profile. I found on some hardware that this parameter value fluctuates. On one of my PCs, booting from CD, ‘eisa:’ always returned the same value, but after a frugal install it fluctuated. I thought perhaps it is a virtual address, but on my laptop it contains non-hex alpha digits — and also on my laptop I don’t have this bug.

Anyway, I modified /usr/sbin/xorgwizard:

#PROFILEMONITOR="`grep -E -m1 '^monitorrange: |^monitorid: |^eisa: ' /tmp/ddcprobe.txt | cut -f 2 -d ':' | tr -d ' ' | sed -e 's/[^0-9a-zA-Z]/_/g'`"
#v3.95 'eisa:' and 'id:' return fluctuating values on some PCs, so search in this order...
PROFILEMONITOR="`grep '^monitorid: ' /tmp/ddcprobe.txt | cut -f 2 -d ':' | tr -d ' ' | sed -e 's/[^0-9a-zA-Z]/_/g'`"
[ "$PROFILEMONITOR" = "" ] && PROFILEMONITOR="`grep '^monitorname: ' /tmp/ddcprobe.txt | cut -f 2 -d ':' | tr -d ' ' | sed -e 's/[^0-9a-zA-Z]/_/g'`"
[ "$PROFILEMONITOR" = "" ] && PROFILEMONITOR="`grep '^monitorrange: ' /tmp/ddcprobe.txt | cut -f 2 -d ':' | tr -d ' ' | sed -e 's/[^0-9a-zA-Z]/_/g'`"
[ "$PROFILEMONITOR" = "" ] && PROFILEMONITOR="`grep '^dtiming: ' /tmp/ddcprobe.txt | tail -n 1 | cut -f 2 -d ':' | tr -d ' ' | sed -e 's/[^0-9a-zA-Z]/_/g'`"
[ "$PROFILEMONITOR" = "" ] && PROFILEMONITOR="`grep '^manufacture: ' /tmp/ddcprobe.txt | tail -n 1 | cut -f 2 -d ':' | tr -d ' ' | sed -e 's/[^0-9a-zA-Z]/_/g'`"
[ "$PROFILEMONITOR" = "" ] && PROFILEMONITOR="`grep '^eisa: ' /tmp/ddcprobe.txt | cut -f 2 -d ':' | tr -d ' ' | sed -e 's/[^0-9a-zA-Z]/_/g'`"

The first commented-out line shows what it was before. I needed to prioritise the choices, not just choose whatever comes first in the ddcprobe output, so I chosen what seems like a suitable order of preference. I left ‘eisa:’ as the very last choice.

Note, this code also has to go into /usr/X11R7/bin/xwin, as that generates the current hardware profile then looks to see if a suitable xorg.conf already exists.


One Response to “Xorg Wizard rerun-at-every-boot bugfix”

  1. BarryK Says:
    January 8th, 2008 at 2:09 pm

    Um, perhaps that could be simpler. How does the ‘-m1′ option to grep work? In this case:
    grep -E -m1 ‘^monitorrange: |^monitorid: |^eisa: ‘
    does it check for ‘monitorrange’ first and stop if a hit, then ‘monitorid’, etc? If so, then I just need to fix that list rather than my long-winded solution.

Universal Installer crashing bug fixed

January 8th, 2008

The Universal Installer in the Dingo alpha4 “full” build has a couple of new bugs.

One reported bug was that it crashed. I found a syntax error and that is now fixed.

However, there was another report, something to do with installing to USB I think, something about an operation that was supposed to happen but didn’t. I don’t know what that problem is, and I’ll need more details to know where to look in the script. Like, exactly what installing to, what step of the install process, what operation did not succeed.

Note, the problems with frugal install, outstanding from earlier alpha testing, are next on my list.


2 Responses to “Universal Installer crashing bug fixed”

  1. Sage Says:
    January 8th, 2008 at 3:34 pm

    Has the disappeared IDE HD issue been solved? I confirmed it on seven more systems, including both cpu brands. Although I cannot boot USB sticks on any of my AMD nor Intel boards, until Dingo, I could install it OK. Now the UI just drops out after the first screen when attempting this.

  2. JustGreg Says:
    January 9th, 2008 at 3:12 am

    Barry, here is what happens when one tries to install Dingo alpha4 to a USB device. I used ps-a to check the process to see where it fails.
    Starting the universal installer results in a new process called puppyinstaller. After selecting usbflash, the puppyinstaller is still there along with one for GTK. Selecting the device (sd?1), the processes are still going, but process numbers do change. Agreeing to the device selected, results in the processes still going. The processes still are working after one picks the CD for the source. The proper device for the CD is listed. Upon clicking on OK to start copying, the CDROM starts and then stops. A check of process shows all process for the universal installer have stopped and are not listed. I did check and found the cdrom was mounted in mnt/sr0. I had made sure the CDROm was not mounted at the start of this. I looked at the puppyinstaller script and one selects the Cdrom in line 699. The error has to be just after that line. I hope this helps.

Async-scsi problem again

January 7th, 2008

Raffy, JustGreg and dvw86 reported “pup_394.sfs not found” when they boot off a USB drive, testing Dingo alpha4 “full” build.

I tried it myself, installed to USB Flash drive, booted, rebooted and created a pup_save at the shutdown, it rebooted fine. Tried another reboot, no problem. So, I did a complete poweroff. Powered up, and this time “pup_394.sfs not found” error.

I suspected the sync-scsi problem that I have discussed in a recent post. In alpha4 ‘barebones’ I had placed a ’sleep 2′ after detecting the usb drives in the ‘init’ script, due to the asynchronous assignment of drive letters to usb drives. However in the ‘full’ alpha4 I had loaded the ’scsi_wait_scan’ module, which is supposed to fix that problem (thus I removed the ’sleep 2′.

Yeah, well it seems not, not properly anyway. I put in a ’sleep 1′ and now the ‘pup_394.sfs’ is getting found reliably. I tested on the Classmate also, and that is ok.

It seems also that the Universal Installer in alpha4 ‘full’ build has problems, so best to give that iso a miss. Next thing I’ll do is find out what’s wrong with the Uni Installer, and there will be another alpha soon.

Pure FTP server, Podcast grabber updates

January 7th, 2008

Kirk has posted about these upgrades in the Dingo-alpha3-feedback forum thread:

http://www.murga-linux.com/puppy/viewtopic.php?t=24534&start=90

I have put the Puppy Podcast Grabber into the official PET repository. This was originally created by Brad_Chuck and updated by kirk.

One question though: It has a binary utility ‘id3tool’. Dingo already has various id3 utilities, in the ‘id3lib’ package. These are ‘id3convert’, ‘id3cp’, ‘id3nfo’ and ‘id3tag’, in /usr/bin. They might not have made it into alpha3, but are in alpha4. They are used by the Pmetatagger package by plinej. Anyway, couldn’t these be used instead of ‘id3tool’? (or vice-versa, if id3tool is better)

Regarding the Pure-FTP server, I’m trying to recall what the history was. I think there were two different packages developed? I took a very quick look at the package in Puppy 3.01, the ‘run-pureftpd’ script was written by David Tangye. It looks quite sophisticated, with automatic creation of users and groups. What actually is wrong with it, apart from the need to launch ‘pure-ftpd’ with ‘-H’ option? Or, is there no need for that extra sophistication?


7 Responses to “Pure FTP server, Podcast grabber updates”

  1. kirk Says:
    January 7th, 2008 at 10:36 pm

    I didn’t know Dingo had those id3 tools. I’ll update the package, hopfully tonight, if I don’t have to work.

    Besides needing the -H option, the run-pureftpd script was over complicated. If Dingo had a user named ftp, with no password, and a home directory, that takes care of most of the complication. Also got rid of most of the options to make it less confusing. The thinking is that someone who wants to use all the different options will probably have no problem running it from command line. But a novice user, who just wants to share files would be better off with a less complicated GUI.

  2. kirk Says:
    January 8th, 2008 at 12:11 am

    Updated the Podcast grabber script to use id3tag.

    http://www.murga-linux.com/puppy/viewtopic.php?p=164649#164649

  3. BarryK Says:
    January 8th, 2008 at 7:37 am

    Thanks for the podcast grabber update, and the explanation about the pure-ftpd script.

    Now, I’m real lazy to request this, but would you mind posting the /etc/shadow, /etc/passwd, /etc/gshadow content to show the entry for ‘ftp’. Um, is a group entry required, /etc/group?
    I guess I’ll need to manually create /root/ftp too.

  4. BarryK Says:
    January 8th, 2008 at 7:52 am

    Just a thought. Would it avoid confusion if the directory was called /root/ftpd rather than just ‘ftp’? There is already a /root/ghttpd, and the ‘d’ on the end implies that this is a directory for a server.
    Otherwise, perhaps if the name was ‘ftp’ then users might think that’s a place to download files to when using the gFTP client application?

  5. kirk Says:
    January 8th, 2008 at 10:43 am

    No problem, used /root/ftpd as the home directory:

    /etc/shadow — ftp:$1$/ZvMubxC$OcCeFwKThRKEPWN6Z1dTK0:13885:0:99999:7:::
    /etc/passwd — ftp:x:1000:1000:Linux User,,,:/root/ftpd:/bin/sh
    /etc/gshadow — ftp:!::
    /etc/group — ftp:x:1000:

    I reposted the tar ball with everything here:

    http://www.murga-linux.com/puppy/viewtopic.php?p=164656#164656

  6. BarryK Says:
    January 8th, 2008 at 5:38 pm

    Okay, I’ve made pure-ftpd into an official PET package for Dingo.

    One detail though. I’m stretching my memory a bit, but wouldn’t this also suffice:

    /etc/shadow:
    ftp:!:13885:0:99999:7:::
    /etc/gshadow:
    ftp:::
    /etc/passwd:
    ftp:x:1000:1000:Linux User,,,:/root/ftpd:/bin/sh
    /etc/group:
    ftp:x:1000:

  7. kirk Says:
    January 8th, 2008 at 10:12 pm

    Now your just showing off! :) Yes, that works. Without looking it up, I wouldn’t remember how to do that manually. I was just giving you the output from adduser.

Fix for Network Wizard

January 7th, 2008

Forum member ‘rerwin’ has fixed a problem with the latest dhcpcd program in Puppy Dingo. This is excellent detective work, applying a fix that handles older dhcpcd’s also. I’ve applied the fix to the Network Wizard, and also updated /etc/rc.d/rc.network. Rerwin’s forum thread:

http://www.murga-linux.com/puppy/viewtopic.php?t=25098

There was a recent upgrade to the Network Wizard, see a little bit down this blog, and this latest brings the package to ‘net_setup-3.96′. If anyone else wants to work on the Network Wizard, please be sure to get this very latest version, so we don’t run into synchronisation problems. It will be in the next Dingo release, and I’ll also update all the PET packages on ibiblio.org — probably in 2-3 days.


3 Responses to “Fix for Network Wizard”

  1. MarkUlrich Says:
    January 9th, 2008 at 4:56 am

    I’m currently localizing this version of net-setup.sh.
    Should be finished in 1-2 days, I reached now line 730 of 1400.

    It puts all strings in external .mo files (english and german).

    If someone already works on updates, please contact me, so I can provide you the current snapshot.
    This is a painfull work, and we should avoid to work on different versions.

    Mark

  2. MarkUlrich Says:
    January 9th, 2008 at 8:34 am

    finished!

    http://puppyfiles.org/stuff/Network-wizard-3.96-locale-update.tar.gz

    Includes:
    Network-wizard-3.96-locale-update/usr/sbin/net-setup.sh
    Network-wizard-3.96-locale-update/usr/share/locale/de/LC_MESSAGES/net-setup.mo
    Network-wizard-3.96-locale-update/usr/share/locale/en/LC_MESSAGES/net-setup.mo

    Use the english .mo as a template for other languages :)

    Mark

  3. MarkUlrich Says:
    January 10th, 2008 at 11:26 pm

    There was a wrong translation of “NOINTERFACES” , the english was in german and vice-versa.
    I updated the archive.

    Thanks to ricstef for reporting it :-)

    Mark

Openbox, Obconf, startup-notification

January 7th, 2008

I have added these to the official PET packages. Openbox is a window manager and Obconf is a configuration GUI. Startup-notification is a dependency library. I had compiled these a few weeks ago when I was testing MPlayer on different window managers. Plinej has also compiled them recently, see his forum thread:

http://www.murga-linux.com/puppy/viewtopic.php?t=25185

Openbox does not have any kind of tray, so it can be used with Lxpanel or Fbpanel (see recent blog post).

MTR: traceroute GUI

January 6th, 2008

MTR is a GTK2 GUI traceroute application, very small so I have put it in Dingo — it’s very interesting to see the tortured path my packets follow with my satellite Internet connection!

Thanks to forum member ‘paulh177′ who posted about MTR. The project author is Matt Kimball and the homepage is:

http://www.bitwizard.nl/mtr/

Lxpanel for Dingo

January 5th, 2008

Plinej has created a PET package for Lxpanel. See forum thread:

http://www.murga-linux.com/puppy/viewtopic.php?t=25118

I have integrated this into Dingo in the same way as for Fbpanel.

It would be nice to restart Lxpanel (and Fbpanel) when the menu is re-generated, but I don’t want it to start in Unleashed or in rc.update — in that case only the menu is to be generated. So, I changed the end of ‘lxpanel_menu_refresh’ (and ‘fbpanel_menu_refresh’) to this:

if [ “`pidof lxpanel`” != “” ];then
killall lxpanel
lxpanel &
fi

In Unleashed, the fixmenus script is run in the ‘rootfs-complete’ folder in a chroot situation, so the pidof would not give a false-hit if the host system is running lxpanel.


3 Responses to “Lxpanel for Dingo”

  1. plinej Says:
    January 6th, 2008 at 4:04 pm

    Good idea. Should’ve thought about that when I made the script. I just modified both packages and included your suggestion as well as fixing some other things.

  2. BarryK Says:
    January 6th, 2008 at 7:43 pm

    Ok, I’ve update to your latest (6th Jan) lxpanel and fbpanel PET pkgs.

  3. BarryK Says:
    January 7th, 2008 at 8:59 am

    A clarification for everybody: I did not mean to imply that Fbpanel or Lxpanel will be in the next release of Dingo, just that they are added to the official PET repository.

Fbpanel in Dingo

January 5th, 2008

Plinej has created a PET package for Fbpanel, a tray/taskbar/menu panel. He has also created the tools to generate a correct menu for Puppy. Forum thread:

http://www.murga-linux.com/puppy/viewtopic.php?t=25103

I have put this package into Dingo, but made some small changes so that it will integrate nicely:

  1. I changed Jason’s /usr/bin/fbpanel_menu_refresh script so that it does not start fbpanel after generating the menu.
  2. The ‘xdg_puppy’ PET package has /usr/sbin/fixmenus, a script that is called by the ‘createpuppy’ script in Unleashed when building Puppy, by ‘rc.update’ when a Puppy-version upgrade or a sfs-layer change, and by PETget when a package is installed. I added a line in ‘fixmenus’:
    [ -f /usr/bin/fbpanel_menu_refresh ] && fbpanel_menu_refresh
  3. /root/.xinitrc is a script that starts various X applications such as the window manager. I added code to start fbpanel:
    #only launch tray for w.m. without inbuilt tray…
    if [ “$CURRENTWM” != “jwm” -a “$CURRENTWM” != “icewm” ];then
    [ -f /usr/bin/fbpanel ] && fbpanel &
    fi

The above seems to have covered all the bases, but I haven’t actually tested it yet.


6 Responses to “Fbpanel in Dingo”

  1. magerlab Says:
    January 6th, 2008 at 1:15 am

    Barry , why not to add fbxkb to puppy ?
    its a small applet that shows keyboard layout from xorg.conf
    http://www.murga-linux.com/puppy/viewtopic.php?search_id=1849232149&t=24274
    hereś a link to the forum thread

  2. BarryK Says:
    January 6th, 2008 at 5:27 am

    magerlab,
    Okay, I’ll check it out. That’s interesting, I see in the link that MU has created /root/autostart for muppy.

  3. MarkUlrich Says:
    January 6th, 2008 at 8:26 am

    yes, it is a small autostart roxapp.
    So you add the roxapp to your desktop, then you can drag files on it.
    Not only executables, but also documents.

    http://murga-linux.com/puppy/viewtopic.php?t=6084

    As dotpups.de is down, get the mentioned files here:
    http://www.puppyfiles.ca/dotpupsde/dotpups/Desktop-Tools/Autostarter-MU.pup

    Mark

  4. pakt Says:
    January 6th, 2008 at 3:43 pm

    Pup214R has fbxkb, set up for choosing between the chosen keyboard and the US keyboard. The applet shows as a tiny flag in the taskbar. Click on it and it toggles to the other flag indicating the changed keyboard layout. Works great. Take a look at it if you want to see how it works in Puppy ;-)

  5. Dougal Says:
    January 6th, 2008 at 9:53 pm

    I don’t mean to sound negative, but what do you need fbpanel/Lxpanel for in Dingo?
    The reason Jason has then is since he uses a WM (pekwm) that has no panel/taskbar… Will Openbox be included in dingo as an alternative to JWM?

  6. BarryK Says:
    January 7th, 2008 at 8:09 am

    Fbpanel/Lxpanel has been added to the official PET packages to provide choices for Unleashed or remastering Puppy builders. Also Blackbox and Openbox are PET packages.

    A clarification for everybody: I did not mean to imply that Fbpanel or Lxpanel will be in the next release of Dingo, just that they are added to the official PET repository.

Pburn updated to v 0.8

January 5th, 2008

Zigbert has developed Pburn, a CD/DVD burner application. It has a very nice user interface, very straightforward to use. I have updated to the latest version. If you would like to try it right now, here is the forum thread:

http://www.murga-linux.com/puppy/viewtopic.php?t=23881

UPDATE, 7th Jan 2008:
I have updated Pburn to v 0.9.0

Pmetatagger updated to v1.1

January 5th, 2008

Plinej has upgraded his Pmetatagger audio file tag editor GUI application to v1.1. See forum thread:

http://www.murga-linux.com/puppy/viewtopic.php?t=24291

mtPaint upgraded to v3.20

January 5th, 2008

mtPaint is a fantastic small paint application created by Mark Tyler and now maintained by Dmitry Groshev. I have upgraded the PET package to v3.20. Find out more at the home site:

http://mtpaint.sourceforge.net/

Pctorrent, ctorrent updated

January 5th, 2008

Jason (’plinej’ on the forum) has upgraded his Pctorrent GUI frontend to version 1.4 and ctorrent to version dnh3.2 with a patch. I have upgraded the official PETs accordingly. See forum thread for further info:

http://www.murga-linux.com/puppy/viewtopic.php?t=14954

Dmidecode

January 5th, 2008

Puppy now has ‘dmidecode’ which reads information from the BIOS. This is handy for identifying a particular make and model of computer, and I have used it to identify the Asus EeePC laptop. Note, the Classmate PC laptop has a BIOS that has not been customised with make/model information, so I had to use other methods to identify it. The reason that I needed to identify these two laptops is they have an 800×480 LCD screen and the usual ‘ddcprobe’ does not return the native screen resoluton.

I’ve uploaded another snapshot of Puppy, still classified as ‘alpha4′ but it has various incremental improvements over the original ‘barebones’ alpha4. Apart from automatic recognition of the Classmate and EeePC screens, the Universal Installer is also enhanced with an installation choice to suit the Classmate.
This build, dated 20080104, has the ’standard’ package selection, with SeaMonkey, and also the full  ‘zdrv’ file. It can be found at:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/test/puppy-4.00-dingo-alpha4/

Mostly I uploaded this as one of the Intel engineers wants to test the install to the Classmate.


9 Responses to “Dmidecode”

  1. wow Says:
    January 5th, 2008 at 3:42 pm

    Bad news:

    Intel quits One Laptop Per Child program
    http://news.yahoo.com/s/ap/20080104/ap_on_hi_te/intel_one_laptop_per_child

  2. Sage Says:
    January 5th, 2008 at 5:10 pm

    Magic! Most of my systems can again see and use this release. However - strangely, neither cfdisk, GParted nor PMount can see a pile of old network IDE drives that I’m busily reformatting for distro testing. The ogre’s OS has no such problem. Please Barry - talk to Jesse, ask him to do the port and recompile or w.h.y. and give us back MUT: it always was far and away more comprehensive and reliable.

  3. BarryK Says:
    January 5th, 2008 at 5:25 pm

    Sage, if even cfdisk and GParted can’t see the IDE drives, MUT won’t help you. It may have something to do with the new libata drivers for IDE drives.

  4. Sage Says:
    January 5th, 2008 at 5:36 pm

    MUT - maybe; you’re the expert.
    I’ve just checked some more IDE discs. None are visible. 3.01 has no such problem, at least, of this nature! Obviously a show-stopper. If there’s a simple patch I could do that, but, otherwise, I’d have to abandon interest again for the duration. Sorry.

  5. BarryK Says:
    January 5th, 2008 at 6:25 pm

    I presume that this invisibility is all on the one PC?
    The libata drivers are probably not recognising that particular hardware — is it very old?

    You can type this in a terminal:
    # cat /proc/partitions
    to see exactly what drives and partitions the kernel is recognising.

  6. BarryK Says:
    January 5th, 2008 at 6:26 pm

    Lobster has started a forum thread to discuss this new “full” alpha4:
    http://www.murga-linux.com/puppy/viewtopic.php?t=25141

  7. Sage Says:
    January 5th, 2008 at 6:42 pm

    Late SktA, 2GHz - should be adequate.
    OP of cat on Lob’s thread.
    Seems I’m not alone - Firefox…….
    Expect a deluge.

  8. Dougal Says:
    January 6th, 2008 at 9:16 pm

    RE: dmidecode and the Classmate (again)

    I’m still not sure about the lack of info: does it say nothing at all?
    The EeePC is not actually identified that way — the places where it says “EeePC” are the serial numbers, where it has “EeePC-0123456789″… seems like they just put that in for lack of any other idea.
    The way the EeePC is identified is “ASUSTeK Computer INC.”, model “701″ — maybe the Classmate has also got soemthing as cryptic?
    The vendor and model info is retrieved thus:
    dmidecode –string system-manufacturer
    dmidecode –string system-product-name
    dmidecode –string system-version

  9. BarryK Says:
    January 7th, 2008 at 8:10 am

    Vendor/Model entries in the Classmate are empty.

Suggested project: ‘a2ps’ equivalent

January 4th, 2008

I recently suggested a project to implement “kernel mode” PPPOE using only the ‘ppp’ package (no Roaring Penguin package), see an earlier blog post. But no bites yet. Note, Dingo does not have the Roaring Penguin package, as it has a Tcl/Tk script for the GUI (but it will be a PET package).

Puppy is basically oriented for printing Postscript files, but the /etc/cups/cupsd.conf can I think (or some other config file) be setup so that CUPS will call a certain program to convert any other file format to Postscript. The package ‘a2ps’ is an example, an “anything to Postscript” converter, however we don’t need anything so big as Puppy already has all converters needed I think.

Puppy has ‘nenscript’, which is a basic text-to-Postscript converter, and for other text files like .doc and .rtf, Abiword can be used in commandline mode to convert to Postscript. I think the ‘netpbm’ package has a utility for Postscript output also so any image can be converted.

The proposed project is to implement a ‘a2ps’ script that utilises all this conversion stuff that is already in Puppy. This script will accept the commandline options that CUPS expects and CUPS will be setup to automatically use this script for non-Postscript files.

I’m posting this suggestion here in case there is anyone looking for some way to get involved in Puppy development. You would need to have a basic knowledge of Bash scripting and preferably some knowledge about CUPS or at least some interest in the printing side of things.

Note, a great place to get some ideas is /usr/sbin/puppypdf, a anything-to-PDF converter written by ‘thoughtjourney’ back in mid-2005. That script shows how Abiword can be used in commandline-mode.

In fact, that would be another little project. We haven’t looked at ‘puppyPDF’ (see ‘Document’ menu) for a long time and perhaps it needs to be updated, expanded or bugfixed. I did use it several months ago, and I recall that it wasn’t clear just what input file formats it supports. Maybe we could add support for images, even SVG?


One Response to “Suggested project: ‘a2ps’ equivalent”

  1. jcoder24 Says:
    January 4th, 2008 at 11:53 am

    Sounds like you are talking about a CUPS-PS version of CUPS-PDF :-D

    Most applications tend to send jobs to the printer in either postscript or text format. For some non-postscript jobs, cups uses filters located in /usr/lib/cups/filter to convert them to postscript. For the jobs not covered by those in /usr/lib/cups/filter a2ps is called automatically via foomatic.

    Therefore we shouldn’t need to do anything other than to create an a2ps. However, we should only need such a filter if we are going to be printing ‘non-filtered’ document types from the console.

    With respect to puppyPDF, I think this has been superseded by CUPS-PDF. puppyPDF is limited to file formats supported by abiword while CUPS-PDF is limited only by an application’s ability to print. For more information on CUPS-PDF see pdf-writer.txt included in the pdf printer thread.

Indian adventure

January 2nd, 2008

I counted my sheckles and decided yes, I can do this. I’m going to India for four weeks. What decided me is there is a new discount airline based in Singapore that offers a very low price between Australia and India — the various levies and airport taxes are more than the actual airfares. All kinds of restrictions, like you don’t get any meals, only 15Kg luggage, probably older planes too. It’s Tiger Airways, partly owned by Singapore Airlines.

I depart on the 5th of February from Perth and change planes at Singapore, at the Budget Terminal of Changi Airport — wow, I’ve been to Changi Airport plenty of times, never knew they had a “Budget Terminal”. I’m returning on 5th March.

I’ve got some relatives-by-marriage in India, but they are in Delhi and I’m going to Chennai, or rather near Chennai (madras). I last went to India about 12 years ago, and always wanted to go back to that area — any Puppy-fans in that neck of the woods?

I’m planning to take one of the Classmate laptops with me — it fits nicely in my carry-on bag. In fact, I’ll travel light, no check-in luggage at all. I’ll keep working on the Puppy project while I’m over there — I was thinking maybe I’ll work on my web pages. Getting online, well I suppose a lot has changed in India since I was there last. There are probably Internet cafes all over the place.


23 Responses to “Indian adventure”

  1. Sage Says:
    January 2nd, 2008 at 8:36 pm

    Fantastic - masala dosai for breakfast! Presumably, the extra fortnight allows for an alien abduction en route…….
    One of the expert coders from the Forum lives around Madras? (think he returned to medicine? Can’t remember his handle.)

  2. lobster Says:
    January 2nd, 2008 at 8:40 pm

    Good News :)

    I should think internet cafes and maybe even a wifi hot spot will be available in Madras. Have a great time. Share a few pics :)

    Check out frappr bottom of this page for Indian Puppys
    http://tmxxine.com/wik/wikka.php?wakka=PuppyLinux

  3. BarryK Says:
    January 2nd, 2008 at 8:44 pm

    Here are some pics of the Budget Terminal:
    http://www.jeffooi.com/2006/05/heres_what_singapores_budget_t.php
    Looks alright — especially the free Internet access. Interesting that there’s no departure lounges.

    When I booked online, I had the option of pre-allocating my seats, at a cost of AU 5 dollars for each flight, or AU 25 dollars for premium seating positions. I declined, as I have always just done the seat allocation at check-in. I wonder though, with a cheap airline would they overbook? — then having a seat in advance would be a definite advantage!

  4. BarryK Says:
    January 2nd, 2008 at 8:49 pm

    Sage, ahem, yes, it’s four weeks. I’ve corrected the post.

  5. charnisingh Says:
    January 2nd, 2008 at 9:11 pm

    I am staying at calcutta. India. if you happen to be here, you are most wellcome to stay with us.

  6. Sage Says:
    January 2nd, 2008 at 9:22 pm

    All airlines overbook. Provided you turn up on time and comply with all their small print, if you are denied embarkation you can sting them for compensation, a free hotel room overnight, meals and a flight next day - at least, that’s the law (internationally); all you have to do is make it stick and that can involve being extremely firm with them! I once stayed in a 5* overnight in Texas - the normal roomrate was $400, but I sneaked a look at the account they sent to Delta Airlines; it had been charged out at $5, and that included dinner + breakfast. Alternatively, if you accept voluntary offloading, insist on a next-day substitution flight, an hotel room AND your $200 disturbance compensation (or an open ticket for the future).
    India is full of expert IT folk, many located in Chennai; it’s where many of the much-maligned call-centres are based. Maybe a chance to visit Bollywood - largest movie studios in the world, bar none. India has more well-educated, middle class folk (~300m) than all souls in the USA + UK together, and more than half of ours wouldn’t qualify for such a description.

  7. Dougal Says:
    January 2nd, 2008 at 9:26 pm

    We used to have BombaiRockers on the forum… last I remember him he finished his medical studies and said he’ll be too busy for Puppy now.
    There’s also the guy with a picture of a cow as his avatar… I think he might be from India, too.

  8. nic2109 Says:
    January 2nd, 2008 at 9:40 pm

    Dougal said : “There’s also the guy with a picture of a cow as his avatar… I think he might be from India, too.”

    That might be forum member ‘Raman’ from “A place where cows are sacred” - which I have always taken to be India.

    Chennai is not only full of call centres, but many of the global IT players such as Dell, IBM and EDS have development establishments there too.

  9. raffy Says:
    January 2nd, 2008 at 10:38 pm

    Yes, Bombayrockers and Raman should be from India. Chennai hosts Novatium.com, aspiring to be a leader in affordable home computing. The data-processing giants are in Bangalore, such as Infosys and Oracle. It’s interesting that India opted out of the OLPC project. Maybe they have aces up their sleeve?

    Just PM if you plan a little side-trip to Singapore. Some friends could assist you there.

  10. raffy Says:
    January 2nd, 2008 at 10:47 pm

    Looking at the pictures: the cafe with shops and Internet - that’s the departure lounge. It’s really not that crowded, even for a budget terminal. You can have tea with milk for S$0.90 and the popular leaf-wrapped Indonesian meal for S$1.50, unbelievably priced goodies inside an air terminal.

  11. cthisbear Says:
    January 3rd, 2008 at 6:15 am

    ” India is full of expert IT folk, many located in Chennai; it’s where many of the much-maligned call-centres are based. ”

    Sage that’s Barrys’ hidden agenda for visiting India….payback for all
    those phone calls to Penjorie.
    We’re with you BK……4 weeks won’t be enough time to complain to them all…ha! ha!

    Anyway Barry I hope you enjoy your trip.
    Maybe someone from Intel India might chauffer you around for a day.

    http://www.intel.com/intel/location/india.htm?iid=intelww+office_india

    Anyway best wishes to you in 2008.

    Chris

  12. BarryK Says:
    January 3rd, 2008 at 4:16 pm

    Charnisingh, thanks for the kind offer. I won’t be going that far, staying in Tamil Nadu.

    Raffy, your report on prices indicates first hand experience? I’ll have several hours to kill while waiting for the connecting flight — are there any particular inconveniences for someone wanting to hang around in the Budget Terminal for that long? I don’t want to go out of immigration then in again, but I guess I would have to if I want to go over to the main Changi terminals.

    Well, I decided to pre-allocate my seating. The wonders of online booking, it was a simple matter to make a modification to my reservation. Pre-allocated seating cost me AU 20 dollars plus a “convenience fee” of AU 2.25. I’ve got window seats all the way!

  13. BarryK Says:
    January 3rd, 2008 at 4:22 pm

    Cthisbear,
    Actually, I’ll be giving Intel some good publicity, as I’m taking a Classmate and I’ll probably be reporting on it in this blog, maybe even a photo or two.

  14. raffy Says:
    January 3rd, 2008 at 7:44 pm

    Yes. There are food stalls and lounge areas in the budget terminal, and a few ethernet connections for laptop users. There was no eee yet when I last dropped by (June 2007), so I have no idea if wifi service is available there. But Changi, site of the 2 main terminals, is very near and shuttle bus is available. It could be fun exploring Changi, or even trying the city train service from one of the terminals (the Boonlay route goes through downtown Singapore).

  15. nic2109 Says:
    January 4th, 2008 at 1:53 am

    By co-incidence Tamil Nadu features in The Independent (one of the UK’s “Quality” national daily newspapers) in an article (plus appeal!) about an education charity operating there.

    It’s really quite inspiring and beyond what they aleady do - which is pretty amazing - makes me think what a bunch of OLPC/Classmate/eeePC machines could add to what’s on offer.

    It really makes me wish for a big lever to get hold of and pull that moves governments and NGOs and that helps them open their eyes to what can be done.

    See http://news.independent.co.uk/appeals/indy_appeal/article3303652.ece for the article.

  16. trev Says:
    January 4th, 2008 at 6:19 am

    Shock! More things we have in common! As well as living in WA, I’ve been a puppy luvver :) and blog lurker since ver 1.2 because it got an old Sharp PC-A290 Ultrportable running at blitzkreig speed.
    My wife (of 27 years) grew up in Madras and we’ve visted India, via Madras twice in the past 3 years, most recently in late 2006. May I suggest, if you’ve not already got accomodation, try the YWCA International Guest House (opposite Egmore train station). They don’t do the godbotherer thing and the A/C ensuite rooms are the best value that we’ve found. We’ve also stayed at the Pandian (also near Egmore), it was OK but not as good. The YWCA is on a large, spread out campus which we found to be a really nice refuge from the Madras downtown traffic. They accepted booking from us via the web before we left Perth (some others ask for “demand draft” deposits which are a pain to sort out from W.A.).
    We also did lots of train booking (Rajdani & Shatabdi Express trains) over the web before we left West Oz and found our berths reserved for us all around India without a hint of problems and without paying out to TOUTS, the sometimes perennial scurge of travellers in India. We also used the YWCA in Delhi in 2006 which was also good but a little more expensive than the one in Madras. Email me if you’d like more specific help and I’ll be glad to give you my daytime (I work for the WA govt) contact details!

  17. BarryK Says:
    January 4th, 2008 at 7:36 am

    Trev, yes, quite a coincidence! I have my accomodation organised, but I’ll remember your advise in case I change my plans and need somewhere to stay in Madras city itself. Ah, the trains …quite an experience, hey!

  18. caneri Says:
    January 5th, 2008 at 4:47 am

    Madras curry..oh my!!

    I can just taste it now…and I think after reading your blog…curry for dinner tonight.

    Have a good time and bring some curry back to OZ.

    Eric

  19. raffy Says:
    January 5th, 2008 at 5:06 pm

    Barry, you may want to look at this:
    http://murga-linux.com/puppy/viewtopic.php?p=164798#164798
    It will catch you with 2 comfortable weeks before your trip. :)

  20. shankargopal Says:
    January 5th, 2008 at 7:42 pm

    Hi Barry,
    I’m Indian and while I’m from Tamil Nadu, currently live in Delhi. So likely I won’t catch you when you come. Certainly great to know there’s at least one other Puppian in the country :).

    Pity that you may not be able to come anywhere north of Chennai - while as a good South Indian I have no affection for Delhi, would have been great to meet! Hope next time you can make it this side… I am a fanatic Puppy fan though I’ve hardly been able to contribute to the community, not being a computer person by profession…

  21. shankargopal Says:
    January 5th, 2008 at 7:49 pm

    Just to add to what Trev said, it’s quite easy to book tickets even without using the web, but there are electronic reservation counters where you can book your tickets without any difficulty (except potentially waiting in line for half an hour). And once you have your reservation you don’t have any problems.

  22. prit1 Says:
    February 2nd, 2008 at 5:34 am

    Barry .. I am from India - right now in Los Angeles. But Madras/Chennai is my place. Lots of friends out there. I know that place very well.

    Let me know if you need anything at all. I can get my friends to help.

    Safe Journey!
    Prit

  23. BarryK Says:
    February 2nd, 2008 at 7:03 am

    Prit,
    Thanks very much for the offer!

Minixcal updated

January 2nd, 2008

Lior Tubi wrote this back in April 2006. Minixcal is a replacement for the clock in the tray. It displays the time as per normal, and the date when there is a mouse-over, but click on it brings up a little calendar and a button to launch Ical. This is a GTK1 application, but I have recompiled it with GTK2 and I changed the application-launching button from Ical to Osmo. I changed the PET package from ‘minixcal-1.1′ to ‘minixcal-1.1.1′.

Here is the forum thread where Lior announced Minixcal:

http://www.murga-linux.com/puppy/viewtopic.php?t=7754

Note, you won’t find this thread by using the forum search facility, as it only searches back about one year. Instead, I used google.

Note also, the Dingo alphas do not have Minixcal, just the standard ‘xclock’, but Minixcal will be in the next release of Dingo.


3 Responses to “Minixcal updated”

  1. Dougal Says:
    January 3rd, 2008 at 10:00 pm

    Any chance of uploading the modified source for this and freememapplet to Puptrix? Everything there it timestamped Dec. 22nd…

  2. BarryK Says:
    January 4th, 2008 at 7:28 am

    Dougal, yes, I’ve been accumulating lots of new packages to be uploaded to puptrix. Will do it soon.

  3. ecomoney Says:
    January 4th, 2008 at 10:21 pm

    Does the new clock allow the time to be changed by clicking or right clicking on it? This is the most natural way.

Network Wizard bugfixes

January 1st, 2008

tempestuous posted this:

Barry, nibl reports that a small modification to the wag-profiles.sh script fixes the problem of not being able to use 8-digit WPA passphrases -
http://www.murga-linux.com/puppy/viewtopic.php?p=162684#162684

And Pizzasgood has suggested another Wifi wizard fix –
http://www.murga-linux.com/puppy/viewtopic.php?t=24961

I’ve applied these fixes and the PET package is now ‘net_setup-3.95′ (the previous one on ibiblio is net_setup-3.91).

Kernel boot params

January 1st, 2008

I have implemented ‘pfix=noram’ recognition in the ‘init’ script. This will prevent the pup_xxx.sfs file from being copied into RAM. Puppy will automatically not copy this file into RAM if the PC has less than 256MB physical RAM, but a PC with 256MB RAM is a critical amount and Puppy defaults to copying pup_xxx.sfs into RAM but you might prefer it did not, to get a bit more working-space in RAM at the expense of slightly slower applications startup.

I’m thinking in particular of the Classmate and Eee baby laptops that have 256MB of RAM. Leaving the pup_xxx.sfs on the internal Flash memory and reading from there will cause a slightly higher application startup delay, but overall it may be a good tradeoff.

I have also put a question into the Universal Installer, asking if you want this ‘pfix=noram’ with an explanation of the pros and cons.

The Universal Installer creates a boot parameter ‘ide=nodma’ when Puppy is installed to an internal CF card. However, with the new kernel it seems that this no longer works. I searched with google and found the following information in a file named ‘patch-2.6.23-git9.log’:

libata: Add a drivers/ide style DMA disable

This is useful when debugging, handling problem systems, or for
distributions just to get the system installed so it can be sorted
out later.

This is a bit smarter than the old IDE one and lets you do

libata.dma=0    Disable all PATA DMA like old IDE
libata.dma=1    Disk DMA only
libata.dma=2    ATAPI DMA only
libata.dma=4    CF DMA only

(or combinations thereof - 0,1,3 being the useful ones I suspect)

(I’ve split CF as it seems to be a seperate case of pain and suffering
different to the others and caused by assorted PIO wired adapters etc)

The problem is, I have searched through the 2.6.23.12 source and cannot find anything about this. Is this only for the 2.6.24 kernel? I don’t know, documentation on this does hardly exists. The above extract was dated Oct. 2, 2007, but does not seem to have made it into my kernel source.

Has anyone tried Dingo alpha4 with a IDE Zip or LS-120 drive? Previously Puppy loaded ‘ide-floppy.ko’ however that no longer exists. It appears that the appropriate libata driver has already been configured built-in to the kernel, but once again I can’t find any definite information.


One Response to “Kernel boot params”

  1. JustGreg Says:
    January 2nd, 2008 at 5:42 am

    The Eee PC model 2G has 256 megabytes of ram. The models 4G and 8G come with 512 megabytes of ram. With the models 4G and 8G, the end user can replace the ram and increase it to, I think, 1 gigabyte. I hope this helps.

“Free memory” confusion avoided

December 31st, 2007

There has often been confusion about the value displayed in the free-memory applet in the tray. The reason was that this could be displaying either free space in the ‘pup_save’ file or in the RAM. In the case of booting off a Flash drive, you would normally have a pup_save file on the Flash partition (or use the entire partition) but Puppy works in RAM and saves periodically — hence there are two places that can run out of space, either the pup_save or the RAM working-space.

I decided to simplify things. The free-memory display in the tray now only displays how much free memory is in the pup_save file. This is what I think users logically expect it to be displaying.

For situations where RAM is being used as a working-area, I am still using the top of the screen as a place to display warning messages. If the working-RAM space gets low (we don’t yet have completely 100% flushing of RAM to pup_save) then a message comes up recommending to reboot which will flush the RAM. Similarly, if the pup_save is getting full, a warning comes up with advice to either resize the pup_save file or delete some files.

The multisession CD/DVD is a special case, as the RAM working-space is really all there is. In this case, the tray is displaying free space in RAM. Um, though I suppose it would be possible to read how much space is left on the CD/DVD and display that.

Note, the ‘freememapplet’ in Dingo alpha4 is my original Xlib-based applet. This has very low CPU usage. There was a bug reported, when JWM was restarted then there were two freememapplets displayed — I’ve fixed that. My package is ‘freememapplet_xlib-0.1.1′ and the executable is named ‘freememapplet’.


3 Responses to ““Free memory” confusion avoided”

  1. PaulBx1 Says:
    January 1st, 2008 at 12:55 am

    Barry, it’s a good idea to make more consistent what freememapplet is reporting. I would encourage making it also “read how much space is left on the CD/DVD and display that”, assuming it is not too tedious or time-consuming to do so; otherwise, simply not displaying anything at all for the multisession case. Perhaps the space remaining could simply be stored in PUPSTATE at boot time, to be accessed by freememapplet later?

    One further change might make sense. If you would change the notion that it is displaying “memory” to displaying “space” or “room”, I think more confusion would be removed. To me, memory implies ram. For example, the flyover might say, “Free space remaining in pup_save file”. If you really want to go overboard on that, change the name to “freespaceapplet”!

  2. rarsa Says:
    January 1st, 2008 at 3:34 am

    I believe that when users read “memory” they (we?) think of RAM, when we read “free storage space” we think of file system.

    Maybe the solution to avoid confusions would be to call it “free storage” if you are referring to filesystem even if the filesystem is a RAM disk.

  3. Me Says:
    January 1st, 2008 at 6:43 am

    Here Hair Heir Hare Hear!!!…:)

Puppy “Dingo” on the Classmate

December 30th, 2007

This new web page says it all:

http://www.puppylinux.com/baby-laptops/classmate.htm

As mentioned on the web page, I still have to fix a few things, but hope to get it sorted out for the next release of Dingo …which may be a ‘beta’.


3 Responses to “Puppy “Dingo” on the Classmate”

  1. lobster Says:
    December 30th, 2007 at 9:20 pm

    dugg it here - go visit and digg guys (wide readership)
    http://digg.com/hardware/Puppy_Frolics_in_the_Classroom

    Looking forward to Alpha5/Beta1
    I am running with Alpha4 with Xorg and it feels Beta-ish :)

  2. Sage Says:
    December 31st, 2007 at 12:54 am

    Interesting. Am I seeing things or is the penultimate PUI entry for real?

  3. slvrldy17 Says:
    January 1st, 2008 at 10:11 am

    Have been looking at the ASUS Eee PC as a possible replacement for my old laptop - since like the Classmate it uses SSD/flash memory for storage would the same install routine work for it? Has anyone tried Dingo on an Eee PC yet?

Wet Puppy wallpaper

December 30th, 2007

Heh, heh, forum member ‘doggtoon’ has created a very novel wallpaper creator for Puppy:

http://www.roscoetoon.lunarpages.com/wall_wet/

Also see the forum discussion:

http://www.murga-linux.com/puppy/viewtopic.php?t=24939

Hotplug not as good as before

December 30th, 2007

I have been doing some work on the Universal Installer script. While testing, I reformatted a USB Flash drive from having a MBR and a ext2 partition, to a “superfloppy” without any partitions.

As a side note, the reason I did this is because I can’t get my recently-purchased Kingston DataTraveller 2GB Flash drive to boot on the Intel Classmate laptop. It boots fine on my Acer Aspire laptop. On the Classmate, the LED on the drive flashes at poweron, so it is getting recognised, and the BIOS is set to boot from it, but it just gets ignored and booting proceeds from the internal operating system. I’ve tried all sorts of things, and the superfloppy was my last attempt, which also failed. Note, every configuration I tried, like different MBRs, ext2 or fat16 partitions, superfloppy, all booted on my Acer laptop.

On the other hand, I’ve got an older Astone 512MB Flash drive that does boot on the Classmate. I even tried copying the MBR from the Astone to the Kingston drive, still no go.

Anyway, when I reformatted the Kingston DataTraveller as a superfloppy, /proc/partitions and /sys/block/sdb still showed that it has a partition. Now, I have had this problem in the past, and the ‘probepart’ script fixed this by reading from the drive — the script uses ‘disktype’ to read the drive, but ‘dd’ also worked. Unfortunately, with the 2.6.23.12 kernel this no longer works. Even running Gparted did not trigger the kernel to update its /proc and /sys.

Accessing the drive used to cause a hotplug event which updated /proc and /sys, in the 2.6.21.7 and earlier kernels, but not in the 2.6.23.12 kernel that I’m currently using.

I got /proc and /sys to recognise the superfloppy by rebooting, but I guess just physically unplugging then replugging the drive would also work. However, this is annoying in my ‘puppyinstaller’ script. I will have to put in a message to replug the drive if a reformatting has been performed, but what about drives that are internal and can’t be replugged? Very annoying.

It would be great if there was a little utility program that you could run like ‘rescan /dev/sdb’ and the kernel would then update /dev/sdb in /proc and /sys.


5 Responses to “Hotplug not as good as before”

  1. Me Says:
    December 30th, 2007 at 9:08 am

    Barry, I wunder if this is a BIOS issue if I’m understanding Y’r issue.
    I’ve got an Asus P4P800S SE MutherBoard that sees USB Devices under 530 MB as Removable Disks ( like Floppy ) & Will Boot from it, Otherwise Ov’r 530 MB recognises it as Mass Storage Device & wont :(

  2. C.H.Lee Says:
    December 30th, 2007 at 11:04 am

    Barry , except the MBR ,the heads and sectors/track is very important. Because many BIOS only accept Head=64 ,Sector/Track=32 when boot from USB-disk . The below steps let all my usb flash disks , SD card and FC card boot from USB , hope it can help you and all Puppy user .
    Suppose the USB disk in /dev/sda :
    1. Use cfdisk /dev/sda to see the partitions , remember the original C,H,S data . Then use the “g” command to change the H=64, S=32 .
    2. In cfdisk , create a new FAT16 (06) partitions for SD card , or any partitions you want . Then write the new partitions and exit cfdisk .
    3. Update the MBR , dd if=/usr/lib/syslinux/mbr.bin of=/dev/sda
    4. Format the new partition : mkfs.msdos -v -F16 /dev/sda1
    5. Update the boot record : syslinux /dev/sda1
    6. The last step is mount the /dev/sda1 and copy all files use need .

  3. pakt Says:
    December 30th, 2007 at 3:23 pm

    I tried a Kingston DataTraveller 1GB Flash drive about half-a-year ago on my Dell laptop and couldn’t get it to boot with a partitition. Seems to be one of the very few (only?) Flash drives I tried that wouldn’t work. I returned it to the store and bought a PNY Flash drive instead. I remember helping someone else who had that problem with a Kingston DataTraveller drive and I suggested he return it.

    Never had any problems with PNY (I have several sizes). SanDisk has always worked for me too (several sizes). I’ve also got some other odd brands for testing that boot without problem.

  4. cthisbear Says:
    December 30th, 2007 at 3:47 pm

    http://www.911cd.net/forums//index.php?showtopic=19612&hl=puppy+linux

    Barrymaybe contact the bright sparks at the 911 CD forum.
    Many know and or use Puppy with Ultimate Boot CD.

    Dietmar and jaclaz and or others may well help you out.
    I know its a Windows forum but any knowledge gleaned is
    always welcome.

    Chris.

  5. tgeorg Says:
    December 30th, 2007 at 7:36 pm

    I have several USB Sticks with 2GB (Transcend, Toshiba, Samsung and Trekstore) and had problems to boot from them as they were manufacturer formated. I think the problem is SYSLINUX can`t boot from FAT32 partitions.
    I had to reformat to FAT16.

’scsi_wait_scan’ module

December 29th, 2007

I reported recently in this blog that I have to recompile the kernel, as I had enabled “Asynchronous SCSI scanning” when doing the configure prior to compiling.

Apparently, this feature can speedup booting, but it can also cause a problem if the SCSI drivers are modules, not built into the kernel. The situation with USB drives falls into this category — the drivers are modules, and there is a delay before the drive letters (sda, sdb, etc) are recognised — this is the SCSI layer that is now used for all drives, IDE, SCSI, SATA or USB.

There is some code in the ‘init’ script that loops while waiting for USB drives to register, but with the “async scsi scanning” the loop exits and the script continues with a possibility that the SCSI drives have not yet been assigned, causing a failure in the next stage of the script. This problem occurs with a slower CPU, such as the Classmate. I fixed it temporarily in Barebones-alpha4 by putting in an extra ’sleep 2′.

However, what I found is that when “async scsi scanning” was chosen, it also created a module, ’scsi_wait_scan.ko’, and if this module is loaded then synchronous behaviour is restored. So, I have chosen this path rather than all the trouble of recompiling the kernel.


3 Responses to “’scsi_wait_scan’ module”

  1. PaulBx1 Says:
    January 1st, 2008 at 1:57 am

    Here’s an interesting discussion of booting and probing in parallel:
    http://lca2007.linux.org.au/talk/210.html

    Perhaps it might give a hint or two; it does mention some of your concerns. In Puppy 2.16 gxine gives the audio but not the video for this .ogg file.

  2. BarryK Says:
    January 1st, 2008 at 9:28 am

    .ogg files are only audio aren’t they? I only get audio with the Xine-plugin in Dingo.

  3. cb88 Says:
    January 8th, 2008 at 2:14 am

    .ogg is a container format i believe and can hold audio or video or both similar to .avi also i believe that a common video format for .ogg is theora….

    that is about all I know on the subject


A template for Barebones

December 29th, 2007

While reading Bob’s comments to a recent post, I was reminded of something that I had forgotten to mention. I created Barebones in Unleashed by using a “recipe” file, “template” file, or “patch” file, or whatever it should be called.

Those familiar with Unleashed will know that it has ‘packages.txt’ which lists all the official PET packages with an “on” or “off” against each one. When running ‘createpuppy’ script, a 2-pane window allows modification of this selection, so some packages can be turned off to create a Barebones Puppy.

The thing is, it is the one file ‘packages.txt’, that gets modified, so if you set it up for Barebones, then you would have to make a backup copy with the previous ’standard’ Puppy selection.

I decided to make this easier. It is nicer if ‘packages.txt’ can be left alone at the ’standard’ package selection, and instead I created a little “patch” file, ‘pkgs-barebones.txt’, that is just a list of what to add or remove from packages.txt, For example:

-abiword
-gnumeric
-goffice
-seamonkey
+netsurf
#-pidgin

Just one package name on each line, if preceded with a ‘-’ then that package must not be in Barebones, if preceded with ‘+’ then that package must be in Barebones, any other character in column one then that line is ignored.

‘createpuppy’ now asks if you want to use the Barebones template file, if so it will apply the mods to ‘packages.txt’, but without changing ‘packages.txt’ itself. There are a couple of advantages to this. Firstly, ‘packages.txt’ is left alone as the reference ’standard’ build, and now the requirements for the barebones build are clearly defined in the form of a template file.

I find it quite nice to open the Barebones template file and see at a glance just how it differs from the ’standard’ Puppy.

Similarly, this could be extended to templates for other builds.


3 Responses to “A template for Barebones”

  1. pakt Says:
    December 29th, 2007 at 3:17 pm

    Good news, Barry - exactly what I thought when I used unleashed - that ‘packages.txt’ be left intact to allow for another attempt in case one or more packages were forgotten without having to backup the whole lot :-D

  2. lobster Says:
    December 29th, 2007 at 4:26 pm

    Here is a thought . . .
    With the unleashed CD you send out
    might it be possible to create something like this from Nimblex
    http://custom.nimblex.net/
    However rather than an online version (another possibility)
    The whole thing would be done from the users own computer.

    Puplets for everyone. :)

  3. magerlab Says:
    December 30th, 2007 at 4:18 am

    Hello, Barry!
    Will You uprgrade Seamonkey to 1.1.7 in Dingo?
    the current version in puppy 3.01 does not allow( may be itś a bug) to install new themes and some addons

cdrtools 2.01.01a36 replaces cdrkit

December 28th, 2007

Zigbert reported a problem with cdrkit and I confirmed it, so have moved over to ‘cdrtools’. Get the ‘cdrtools-2.01.01a36′ PET package for dingo from here:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/pet_packages-4/

Zigbert’s ‘Pburn’ is broken in Dingo, and cdrkit was one reason. After substituting cdrtools, Pburn improves but is still broken when I try to burn files to a DVD — fails at the final burn step. We are working on it, see the forum thread. Note, I did manage to burn an iso file to DVD with Pburn. Forum thread:

http://www.murga-linux.com/puppy/viewtopic.php?t=23881


One Response to “cdrtools 2.01.01a36 replaces cdrkit”

  1. plinej Says:
    December 28th, 2007 at 6:55 am

    I’ve burned at least 10 data dvds with Pburn in Dingo w/ cdrkit and haven’t had any problems yet.

Barebones: Boxing Day treat

December 26th, 2007

Boxing Day, the day after Christmas Day. Thought I would upload something for those who get withdrawal symptoms when they don’t have a new puplet to play with.

Barebones is not that bare, with lots of applications. Xvesa only though. No office apps, and only NetSurf for browsing. This is still an ‘alpha’, not yet an official release. The live-CD iso file is only 53.3MB, get it from here:

http://distro.ibiblio.org/pub/linux/distributions/puppylinux/test/puppy-4.00-dingo-alpha4/

I uploaded a heap of PET packages to the ‘pet_packages-4′ folder at ibiblio.org, and the PETget package manager can now be used to install them. So, if anyone wants to play and try a remaster, go for it — I haven’t tested the remaster script in Dingo, so this is an opportunity to work on it and get any wrinkles out.

Note, what I aimed for with this barebones was not to be really bare, but to have the entire infrastructure in place. That is, all libraries, all hardware tools, printing, everything except for the really big applications like office and web browser. So, most of the PET packages will install without needing any extra dependencies.
What about Slackware packages? Yeah, they should work, but some dependencies may be required. Gslapt should work, which would be interesting to experiment with.

Qt4 is a PET package, but I don’t have any Qt4-based PET packages available yet. The Opera developers were planning to release a Qt4-based version when I looked almost two months ago — has that happened yet? — the developers have a prerelease site where advance alpha/beta builds can be obtained.

NetSurf as a web browser… it supports CSS which is most important, but the lack of Javascript is a problem. I was able to login to this blog but the browser didn’t render the toolbar when I wanted to make a post. NetSurf crashed when I clicked on the ‘Sign Out’ button. It’s usable, but only just, and I was pleased to be back on SeaMonkey.

Puppy’s internal Help page is a problem, as it has Javascript for the popup menus. Perhaps we need a non-Javascript alternative. There was a project awhile back I think, for a improved Help page — I think it was in one of the community puplets?

NetSurf for the CUPS web interface? I haven’t tested it recently, but that should be working. The CUPS pages have CSS, which NetSurf can handle.


13 Responses to “Barebones: Boxing Day treat”

  1. lobster Says:
    December 26th, 2007 at 3:57 pm

    “There was a project awhile back I think, for a improved Help page — I think it was in one of the community puplets?”
    Barry I seem to remember this was included in Puppy Viz 2.15CE. It used a two panel system and was very good. Not sure if WhoDo did it or somone else . . . maybe someone can confirm?

    Thanks for this Box Day present gonna try it on an old laptop now :)

  2. cthisbear Says:
    December 26th, 2007 at 4:59 pm

    As posted on forum:

    Hooray she works on NTFS.

    Just cleaned up my XP install with no hiccups.
    Internet working…….but I cannot change the history to zero
    in NetSurf .

    I did not test sound.
    Download….burn CD….boot…delete XP files….browsing the internet
    in ten minutes total.
    Pull up your socks young Barry….not quick enough…..Ha! ha!

    Happy New Year….I hope you enjoy a break soonest.
    Regards Chris.

  3. aran Says:
    December 26th, 2007 at 8:37 pm

    Barry is it also possible to release a new Version of the Puppy Linux Comand Line Only Version called “Puppy One Bone.

    http://puppylinux.org/wikka/OneBone

    The latest Version is from “June 28, 2006″

    It would be great to have a update for that Version.
    The Unleashed Script can’t also not be found any more on the Website or the Net.

  4. Henry Says:
    December 26th, 2007 at 9:14 pm

    Nice surprise, Barry,

    Preliminary impression is very good. Experimental upgrade from Nop 3.01, my current work system, seems OK, apart from package incompatibilities, of course.

    Henry

  5. GreatnessGuru Says:
    December 27th, 2007 at 6:04 am

    Bombed…

    - At the “personal file” chooser I “choosed” a “/pup_save.2fs” file.
    “Version update … 213 to 394″
    That part seemed to go well.
    (I have previously updated another file to Puppy 3.01.)

    “Perform a ’switch_root’ …
    Making … usable…
    Checking … 213 to 394…
    Loading … modules
    puppyserialdetect is running
    Detecting …: ps/2 …: ps/2
    Loading … map…
    Setting … interfaces…
    Detecting … modem… new setup… 1 2 …
    … ttyS_PCTEL0
    Executing … /etc/rc.d/rc.local…
    33[1;31mSorry, cannot start X.
    Link /usr/X11R7/bin/X missing.
    (suggestion: type ‘xorgwizard’ to run the Xorg Video Wizard)
    # _[blinking]”

    All this on an HP Vectra VE (Pentium something-or-other), 64 MB, huge Linux swap partition.

    This may be an improvement of sorts since the 393 (alpha 3) ISO never even got close to this far.

    Thank you,
    Eddie

  6. Dougal Says:
    December 27th, 2007 at 9:18 pm

    Barry, you should report the netsurf problem to the developers, as it’s not javascript-related: I run Firefox with javascript disabled and this page works fine for me…

  7. BarryK Says:
    December 28th, 2007 at 7:31 am

    Pakt has started a forum thread for discussing barebones:
    http://www.murga-linux.com/puppy/viewtopic.php?t=24888

  8. BarryK Says:
    December 28th, 2007 at 8:11 am

    Cthisbear has also started a forum thread:
    http://www.murga-linux.com/puppy/viewtopic.php?t=24863

  9. Bob Says:
    December 29th, 2007 at 3:40 am

    What I would like to see is:

    1) the smallest possible system that can bootstrap itself to a bigger system
    2) a classic Puppy that boots up to X with a carefully selected set of desktop apps, preferably one app per function
    3) recipes to build a custom Puppy for the
    3a) Puppy developer (e.g. pup_394.sfs + zdrv_394.sfs + devx_394.sfs)
    3b) artist (classic + Inkscapelite + …)
    3c) home internet server (dynamic dns + ftpd + httpd + rsyncd)
    3d) system rescue cd
    etc.

    That way, Puppy customizers can concentrate on the recipes instead of ISO’s.
    —
    BTW, running gslapt I get:

    gslapt1: error while loading shared libraries: libgpg-error.so.0: cannot open
    shared object file: No such file or directory

    Thanks.

  10. WhoDo Says:
    December 30th, 2007 at 12:46 pm

    “There was a project awhile back I think, for a improved Help page — I think it was in one of the community puplets?”

    Lobster has it correct. It was in 2.15CE and it was created by headfound as his contribution to the project. I still have it in tar.gz format, if someone wants me to upload it somewhere.

  11. zygo Says:
    January 2nd, 2008 at 1:11 am

    In the help page I have converted all the selection items to link anchors so that when javascript is not available they can be seen. When js is avaialble then it converts the anchors into a selection control as the original help page had. Sadly, http://www.murga-linux.com/puppy/ is all but dead:-
    ——————————-
    Read timeout.
    Line: , Size: 0
    Notice: Undefined index: http_status in /srv/www/web-sniffer.net/html/index.php on line 913
    Notice: Undefined index: HTTP Status Code in /srv/www/web-sniffer.net/html/index.php on line 254
    HTTP Response Header
    Notice: Undefined index: HTTP Status Code in /srv/www/web-sniffer.net/html/index.php on line 273
    ——————————-
    so I have no where to post it.

  12. zygo Says:
    January 2nd, 2008 at 1:20 am

    Or maybe that’s just web-sniffer.net. http://www.murga-linux.com/puppy/ has responded now. It is extremely slow.

  13. zygo Says:
    January 2nd, 2008 at 4:42 am

    It’s uploaded to http://www.murga-linux.com/puppy/viewtopic.php?t=24863&start=28

Osmo now with iCalendar support

December 26th, 2007

I compiled ‘libical’ v0.27 statically, then recompiled Osmo. So, Osmo now has iCalendar compatibility — I don’t yet know exactly what that means (import, export or both).

Patched 2.6.23.12 source for Dingo

December 26th, 2007

In Dingo alpha4 I am using the 2.6.23.12 kernel. I have uploaded the patched source and all third-party drivers to puprix.org, hosted courtesy of Ted Dog:

http://puptrix.org/sources/kernel-2.6.23.12-dingo/

Note, in a day or two I might recompile the kernel, as I had chosen something in the config that I am doubtful about — there was an option for “asynchronous scsi” which was supposed to speedup booting, however I think it caused trouble in the ‘init’ script and I had to put in a couple of seconds extra delay into the USB detection. So, I’ll turn that off, which may mean recompiling the third-party drivers also.

The puptrix site has the ‘.config’ file, so if there’s anything that you think urgently needs to be modified in that file, let me know fairly soon, before I do the next compile. That will be it, the final kernel for Dingo, as it is a lot of work to go through all those third-party drivers, plus I have updated alsa which I didn’t upload to puptrix.

UPDATE: I have uploaded the alsa-drivers source to puptrix. It was a SVN snapshot, dated 2007-12-21.


2 Responses to “Patched 2.6.23.12 source for Dingo”

  1. Flash Says:
    December 26th, 2007 at 9:45 am

    Happy New Year!

    Is this the right time to bring up Alsa drivers for motherboards with on-board HDMI output? (See this thread in the forum:
    http://www.murga-linux.com/puppy/viewtopic.php?p=160016#160016
    for more details.)

    Also, will the new kernel include support for dual-core AMD processors?

  2. BarryK Says:
    December 26th, 2007 at 10:38 am

    I ticked the box for SMP in the kernel config, so I guess that means yes, your dual-core system is supported. I don’t know if anything else has to be done. I left /proc/kcore disabled, don’t know if that matters.

Pbackup, Pmetatagger, id3lib

December 25th, 2007

Package ‘id3lib’ is added, as it is a preferred dependency of Pmetatagger. Version 3.8.3.

Pmetatagger is updated to v1.0.

Pbackup is updated to v3.0.2.

Pburn, Pcdripper, Pmetatagger updated

December 25th, 2007

Just a quick post, I’ve updated these:

Pburn v0.6
Pcdripper v3.2
Pmetatagger v0.9.1

Now I have to tear myself away from this computer and go to some family Christmas Day gatherings.


2 Responses to “Pburn, Pcdripper, Pmetatagger updated”

  1. plinej Says:
    December 25th, 2007 at 10:02 am

    Merry Christmas to you Barry! BTW, pmetatagger is at 1.0 now. If you want to support version 2 tags for mp3 I’d remove mp3info and put in the executables from the id3lib package (id3info, id3convert, & id3tag). I modified how the program works quite a bit in the last release.

  2. mcgregor Says:
    December 25th, 2007 at 1:06 pm

    Merry Christmas!

Osmo personal organiser

December 25th, 2007

This is a great little app, discovered by ‘muggins’. The latest version is 0.1.6 (Puppy 3.01 has v0.1.1) and muggins kindly posted the source to the forum as I have never been able to connect to the Osmo project home page. Here are the links:

http://www.murga-linux.com/puppy/viewtopic.php?t=22917
Osmo home: http://clay.ll.pl/osmo/

One thing I noticed when compiling, it can link with ‘libical’ if available. I haven’t read any of the docs on this, but perhaps that would add the capability to import and/or export from/to Ical organiser files. Would this be desirable to have?


One Response to “Osmo personal organiser”

  1. fwiffo Says:
    December 25th, 2007 at 7:03 pm

    For me: Yes!

    ical is the standard for exchanging calendars over the net. IIRC, this way it would be also possible to export a calendar created with osmo to, e.g., google calendar (and the other way round, too).

    Please correct me if I´m wrong, but if the library isn´t that big or much of a dependency hassle, it would be a nice thing to have.

    Cheers

Automatic detection of 800×480 for Classmate

December 25th, 2007

The LCD monitor in the Intel Classmate PC laptop does not have EDID support, so when Puppy runs ‘ddcprobe’ it returns “edidfail”. So, I needed some way to determine that Puppy is running on a Classmate with 800×480 LCD screen. This is not the best solution, as other tiny or embedded systems may match these specs and have a different screen, but this is what I did:

1. Detect “edidfail” when run ‘ddcprobe’.
2. Detect 915GM/910ML/915MS video.
3. Detect Celeron M 900MHz CPU
4. Detect “scsi 8:0:0:0: Direct-Access M-Sys uDiskOnChip”

The latter is the internal USB 2GB Flash memory. Intel will be bringing out a new model sometime, probably with a bigger screen, so I will have to revisit this test sequence.

Now, Puppy detects the Classmate and knows it is 800×480 only and starts up immediately with that resolution. This works for both Xorg and Xvesa.


4 Responses to “Automatic detection of 800×480 for Classmate”

  1. JustGreg Says:
    December 25th, 2007 at 9:47 am

    The ASUS Eee PC has the same problem. I can force 800 by 480 in Puppy 2.14R and 3.01 with the XVESA server by placing the command “915resolution 41 800 480″ in /etc/rc.d/rc.local I have not be able to do the same with the Xorg. I will say, once Puppy is in the 800 by 480 mode, the screen does look nice. The ddcprobe maybe a hardware issue with the intel graphic chip. Since the intel chip is common to both designs.

  2. Dougal Says:
    December 25th, 2007 at 8:53 pm

    Barry, i was just about to post the simple solution for you here…
    JustGreg sent me the output of “lspci -vv” for the eeePC and I found that nearly all devices had the subvendor/subdevice number 1043:82d8, so I think that can be counted on as identifying the eeePC — the Classmate will probably have something similar.

  3. pakt Says:
    December 26th, 2007 at 4:53 pm

    The BIOS usually contains the PC manufacturer and model name, so these strings could be looked for and used to identify a machine.

    I found this link that may be of use:
    http://freshmeat.net/projects/dmidecode/

  4. Dougal Says:
    December 26th, 2007 at 9:55 pm

    I compiled dmidecode a few months ago, but I think it only works on laptops.

T2 build system for Dingo

December 24th, 2007

Updated web page:

http://www.puppylinux.com/pfs/

Download:

ftp://ibiblio.org/pub/linux/distributions/puppylinux/t2-dingo/

I must emphasise that this is of interest to a very limited number of persons. It is NOT a Puppy distro, it does NOT create Puppy from source. It only builds raw binary packages, which is a long way from a Puppy system.

Therefore, unless you really know that you want T2-Dingo and exactly what for, there is no point in downloading it.

Tch, tch, here I am working on Christmas Eve …if you can call developing Puppy “working”. Merry Xmas everyone!


5 Responses to “T2 build system for Dingo”

  1. raffy Says:
    December 24th, 2007 at 5:17 pm

    Maybe that’s your gift to the super-developers for Puppy Linux. Average contributors like me have already been enjoying lots of fun in the regular releases.

    Cheers!

  2. lobster Says:
    December 24th, 2007 at 7:16 pm

    :)
    Glad you are having fun. I have been busy making a (non) traditional squid and mandarin pasta salad with port in case Rudolf turns up (he got a red nose by drinking port incidentally).
    I am very keen on the T2 system. Probably even more so if I was smart enough to use it. I hope to be allowed to play with the computer on Christmas day, we shall see . . . Wishing you and everyone a Dingo Christmas (I hope that is a good thing - maybe I am thinking of ‘fair dinkum’). Remember some Puppys are not just for Christmas, they are for the whole year. Be Happy. Have Fun. Be Kind.

  3. veronicathecow Says:
    December 24th, 2007 at 10:24 pm

    Hi Barry, just wishing your a very happy Xmas (and a restful one as well, you deserve it)
    Tony

  4. Henry Says:
    December 25th, 2007 at 5:34 am

    Barry and all, Merry Christmas!

    Thanks for a great project, an adventure to behold, valued by us all. 4.0 is a beacon. 3.01 is a sound basis. Two great examples using this are MU’s Muppy and Gray’s NOP. Don’t overlook the latter - it’s the best to efficiently make your own, starting with Xfce, Opera, Xine-gui - and not much else - and filling in your choices!

    Henry

  5. kirk Says:
    December 26th, 2007 at 12:56 am

    Thanks Barry and Merry Christmas! Piglet has been building away for about 8 hours. I had tried T2 a couple months ago, but got hung up with the stage 1 Perl problem.

    Now I feel bad, I didn’t get you anything. :)

MMC/SD card fixed in Dingo

December 24th, 2007

A bug report was that MMC/SD cards no longer work in Dingo. I found two bugs.

Firstly, /etc/modprobe.conf has a couple of lines to load the ‘mmc_block’ module with the parameter ‘major=253′, which is the major device number of /dev/mmcblk0 device names. However, the ‘mmc_block’ module no longer supports any parameters. Consequently, the module failed to load.

It seems that the kernel developers have settled on the major device number of 179 for MMC/SD devices, so the ‘major=’ parameter has become unnecessary. I got the ‘179′ by examining /proc/partitions. This leads to the second bug: I had to recreate the device names:

# mknod /dev/mmcblk0 b 179 0
# mknod /dev/mmcblk0p1 b 179 1
# mknod /dev/mmcblk0p2 b 179 2

I got the SD card out of my camera, popped it into the slot in my laptop, and Pmount recognised and mounted it. Yippee!


One Response to “MMC/SD card fixed in Dingo”

  1. nic2109 Says:
    December 24th, 2007 at 8:26 pm

    Great news, Barry. Thanks.

    Here in the UK it’s just afternoon on Christmas Eve, so that must make it Christmas Day for you, so Merry Christmas to you and all your readers.

    Nick

Puppy 2.16CE

December 24th, 2007

Well, I don’t want this blog to become the place for announcing everyone’s custom puppies, as that means more work for me, and the Community Wiki news page fills that function.

But, as I mentioned 2.14R, I’ll also mention 2.16CE (Community Edition) as another option for those keen on the 2.1x series. Forum member ‘tronkel’ has developed this, and the forum announcement is here:
http://www.murga-linux.com/puppy/viewtopic.php?t=24803

You may have noticed the ‘puppylinux.ca’ URL for some downloads. Eric (’Caneri’ on the forum) has set this up and is kindly providing hosting for people who would otherwise not have anywhere to upload their custom Puppy and/or PET packages and/or .sfs files.

Caneri is also hosting tronkel’s files:
http://www.puppylinux.ca/members/tronkel/


One Response to “Puppy 2.16CE”

  1. raffy Says:
    December 24th, 2007 at 8:58 am

    Thanks, Barry, for taking time to post these developments in your blog. And Merry Christmas to you and everyone!

Muppy Linux

December 23rd, 2007

I have made a couple of announcements recently in this blog about Puppy 2.14R “Revisited”, an older version of Puppy brought up-to-date with many enhancements and back-ports. Scroll down to find out more. Basically though, “Revisited” is for those who like the 2.1x series and/or whose PC can’t handle later kernels.

There are various branches of Puppy that are currently very active, and there was a chart drawn up to show this graphically — a bitmap image I think, so perhaps we need to draw it in InkscapeLite so that we can regularly update it. Anyway, I would like to mention another branch, or what we call a “puplet”, Muppy.

Muppy is based on Puppy v3.01. The “Puppy3″ series is based on Slackware 12 binary packages, so this series is mostly of interest for the compatibility with the large range of Slackware packages available on the Internet. I think also Hacao Linux is based on Puppy3.

Muppy aims to be a more complete solution out-of-the-box, so is a bit bigger than your usual Puppy. But, you get stuff like WINE, Open Office and Java.

So, if you are currently running Puppy 3.0x and interested in upgrading, check out Muppy. Mark Ulrich (MU on the forum), the developer of Muppy, is a very keen and prolific long-term contributor to Puppy, and Muppy is likely to be supported and improved well into the future.

There is a bit of a pattern here. 2.14R for older hardware, Muppy for Slackware12 compatibility and all the “bells and whistles”, and my Puppy4 “Dingo” which is small and bleeding-edge. Well, this could all morph into something else entirely …the Puppy project can never be described as static.

Incidentally, the upgrade path for Puppy is to move to a higher version number. Basically, the higher the version number, the more recent the libraries and packages used. This is a very general statement, but you should be able to upgrade to a higher version number, even if it is from a “revisited” or otherwise enhanced earlier version. For example, you should be able to upgrade from 2.14R to 3.01 to 4.00 (not yet released), but you can’t go back. However, it may be that some of these branches will grow out on their own and offer an alternative upgrade path.

Find out more about Muppy:
http://dotpups.de/Minisys-Linux/Minisys-Linux-en.html
http://puppylinux.org/wikka/MuppyPuppy

And download:
http://www.puppylinux.ca/muppy/
http://sourceforge.net/project/showfiles.php?group_id=141936


2 Responses to “Muppy Linux”

  1. Dougal Says:
    December 23rd, 2007 at 9:18 pm

    There’s a coding error in the init script in 3.93 and 3.01 (which was probably introduced in 3.00).
    This line:
    if [ WAITUSB=”yes” ] ; then
    always returns “true”, since you’re testing a variable assignment, not comparison.

  2. BarryK Says:
    December 24th, 2007 at 6:27 am

    Dougal,
    Oops! Thanks for that, I’ve fixed it.

Universal Installer updated

December 23rd, 2007

Well, I haven’t tested it yet and there might be more to do, but basically I’ve updated /usr/sbin/puppyinstaller for Dingo. There are no longer any /dev/hd* devices.

I thought about making it work with or without the old /dev/hd* devices, but it is too awkward, so have moved into the future and no going back.

There was a bug report about missing ‘mcopy’ when installing to a USB Flash drive, but the script does not use mcopy. I think it’s the ’syslinux’ utility — the syslinux package has two versions of ’syslinux’, one needing the ‘mtools’ package, the other not. It would seem that I had the wrong one in Puppy, that I have now rectified.

gcc bug fixed

December 22nd, 2007

I described this bug in the “Grumble, grumble” post.

Kirk found that after recompiling gcc, the bug went away. I have also confirmed this. I followed the steps from T2 as closely as possible, to make sure that the libraries, such as ‘libstdc++.so.6.0.9′ are as compatible as possible with those created in T2. This involved applying some small patches, and the same configure options as used in T2. For the record, this is what I did:

# patch -p1 < ../fixincl.patch
# patch -p1 < ../libstdcpp-with-tag-cc.patch
# patch -p1 < ../no-install-libiberty.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 –host=i486-t2-linux-gnu –enable-__cxa_atexit –disable-checking –disable-bootstrap –disable-libstdcxx-pch –disable-multilib –enable-languages=c,c++ –disable-libmudflap
# make
# make install

I then successfully compiled ‘gparted’, that had previously failed with that same error.

I have now updated Puppy with this new ‘gcc’, but I also contacted the T2 developers as this is not a good situation, having a gcc that does not work quite right outside of the T2 build environment.


5 Responses to “gcc bug fixed”

  1. kirk Says:
    December 23rd, 2007 at 10:57 am

    Barry,

    Where you going to post your Dingo-T2 build package? Or is it too soon?

  2. crafty Says:
    December 23rd, 2007 at 1:59 pm

    Might be off subject - but you may want to check out this link - for possible progs (tiny/small/medium/large) that could be used in Puppy - that is if you don’t know of it already..!!

    http://fresh.t-systems-sfr.com/linux/src/

    There is also - /unix/src - directory at the same site - could even find some useful progs in there as well..

    Just my lowly 2 c’s worth.

    BTW - Have a merry Christmas - take a break and enjoy yourself this Tuesday..

    Cheers.

    crafty.
    .

  3. BarryK Says:
    December 23rd, 2007 at 4:14 pm

    Kirk,
    I might upload the entire T2-build system complete with all the downloads. It’s about 1GB. I wonder if ibiblio will mind…
    Or, there is Ted Dog’s puptrix site.

  4. planegoofy Says:
    December 23rd, 2007 at 11:12 pm

    Barry,

    Can the T2 build system be distributed as an sfs file? Or do we have to rebuild it? (I apologize for my ignorance with regards to T2.) My other question is can we build additional packages with T2 and create unleashed/pet/sfs packages for Dingo? I am looking at XFCE (which you were already planning on…) KDE and various other packages. Thanks for all of your Excellent work on Puppy and have a wonderful Holiday Season and a Happy New Year.

    Jeff

    p.s. Thanks Kirk, I’ve been meaning to ask this question since te lates I found was for 2.10

  5. kirk Says:
    December 24th, 2007 at 2:44 am

    That sounds great. Looking at what the other distos have hosted at ibiblio, doesn’t look like they would mind.

NoteCase updated

December 22nd, 2007

I have updated NoteCase to version 1.7.4. This is a superb outliner, basically useful for any kind of hierarchical record-keeping. I am now using it for my records on the Puppy project, basically any notes, like how I compiled an application or how I fixed a bug, or contacts.

NoteCase documents have a ‘.ncd’ filename extension, but NoteCase also supports encrypted documents, with ‘.nce’ extension. This works well. I used “Save as…” to save my unencrypted document as an encrypted document, just to try it out. Documents are encrypted with the Blowfish algorithm, so pretty safe, though of course there is the issue of the swap file or swap partition that can leave behind unencrypted information.

I improved mime-type handling for NoteCase documents in ROX-Filer. There is now a nice “suitcase” icon displayed, and encrypted documents are now recognised.

Read more about NoteCase:
http://notecase.sourceforge.net/

Module alias bugfix

December 21st, 2007

In the previous post I described a problem with the Firewall Wizard, that I have now fixed. The problem is that the firewall script tries to load modules that do not exist in recent kernels, or rather they are replaced by a new module name.

The old name is an alias for the actual name, or rather is supposed to be, in the recent kernels. I fixed the firewall script, but it also requires a fix in the /sbin/modprobe script, which I have also done.

Puppy now has /lib/modules/modules.alias.<kernel-version>, which is a file with a list of all the aliases. Alias information is actually stored in each module, and the ‘modinfo’ utility can retrieve it, but when ‘depmod’ is run this information is collected into one file, ‘/lib/modules/<kernel-version>/modules.alias’ — but this will not include aliases of modules that are only on the zdrv file. The full modules.alias file is in the ‘zdrv’ file, but it is the easy way out to also have it in the running Puppy.

The ‘modprobe’ script now reads /lib/modules/modules.alias.<kernel-version> and determines if the module-name is an alias and then obtains the actual module name.

Grumble, grumble…

December 20th, 2007

This is a somewhat negative “progress” report.

I decided that the 2.6.24-rc4 kernel is too immature, so I rolled back to the 2.6.23.12 kernel. I compiled it with basically the same configuration, including SMP and libata handling of IDE drives. This time I also compiled a whole heap of third-party modules:

Modem
536, 537, ltmodem, pctel, slmodem, ungrab-winmodem.
Unfortunately, ess v0.3 does not compile.

Wireless
acx, madwifi, iwlwifi, rt61, rt73, rt2400, rt2500, rt2570, ndiswrapper.
I wanted to compile just the USB module out of linux-wlan-ng but it would not compile.

Sound
sis7019

Now for the negative news…

DHCP bug

Running a full install with this new kernel, I connected to the Internet using my usual DHCP, and that works, I am running it right now. Unfortunately, it causes Rox windows to have a delay of about 4-5 seconds before appearing and SeaMonkey takes about 15 seconds to start. If I do ‘ifconfig eth0 down’ or kill dhcpcd, these apps revert to their normal fast startup (virtually instantaneous for Rox).

I’m just about to see if there’s a later version of ‘dhcpcd’ and try that…

Non-existent function

This one really cheeses me off. I tried to compile RutilT and got the message that ‘__sync_fetch_and_add_4′ function is missing at the final link step.

I have already mentioned this problem, it affects a lot of packages. They can be compiled inside T2, but not in Puppy. I asked Rene, the main T2 developer, and he kind of confirmed the cause of the problem, but I still have no solution.

Note, I’m not the only one with this problem, as a quick google reveals. The problem is in gcc/libstdc++ and is something to do with gcc thinking that the host CPU is a i386, and it then tries to link-in a non-existent function. I found a discussion thread back in 2006 where the gcc developers were discussing this bug, and the sentiment expressed was that this bug should simply not be allowed to occur, and they were looking at various patches.

Anyway, here we are at December 2007, with gcc version 4.2.2 and I’m stuck with this show-stopper bug. For packages that exhibit this problem, I have tried injecting ‘-march=i486′ or ‘-mcpu=i486′ as a C flag, which is a solution suggested at various places on the Internet, but no good.

So, I’m just about to approach the gcc developers directly…


16 Responses to “Grumble, grumble…”

  1. BarryK Says:
    December 20th, 2007 at 7:24 pm

    Nup, I just tried dhcpcd version 3.1.8, the very latest. Still get the slow app startup when the interface is up.

    Note, this PC uses the ’sis900′ ethernet module, and I have a wired ethernet connection to my router/modem (which has a DHCP server) …a very plain-vanilla setup.

    Well, I will now sift through the kernel config, see if there’s anything in there that might seem to be related to this problem.

  2. kirk Says:
    December 21st, 2007 at 5:11 am

    Strange it compiles fine in T2. I wonder if the error I’m getting with glxgears is related:

    glxgears: /usr/lib/libstdc++.so.6: no version information available (required by /usr/X11R7/lib/libGLU.so.1)

    It seems this is caused by using a different version of gcc than what the library was compiled with. Could T2 be building with the gcc that’s in your T2 host disto instead of the one it builds? I thought T2 would chroot in to a build environment that it compiled. Maybe using dingo as the T2 build host would work? Of course I’m a moron :>

    Keep up the good work!

  3. BarryK Says:
    December 21st, 2007 at 7:30 am

    The “slow startup” has gone away. The problem is, I don’t know why.

    Last night I recompiled the kernel twice, with different config options, still got the same slowness.

    Then I realised that ‘fastboot’ was on, turned that off, rebooted, startup-speed back to normal. It seems that some vital module did not load, but unfortunately I did not keep a record of what was loading before.

    I thought that it might have been the firewall. The firewall seems to be working, but a very different set of modules loads. Back on the 2.6.21.7 kernel, I get these modules loading when run the ‘default’ choice in the firewall wizard:

    iptable_mangle iptable_nat ip_nat xt_state ipt_REJECT ip_conntrack_ftp ip_conntrack_irc iptable_filter ip_conntrack conntrack_irc ip_tables

    whereas with my latest 2.6.23.12 kernel:

    iptable_mangle iptable_nat nf_nat nf_conntrack_ipv4 xt_state nf_conntrack nf_netlink ipt_REJECT iptable_filter ip_tables

    I know very little about the netfilter firewall, so have no idea what these differences mean.

  4. BarryK Says:
    December 21st, 2007 at 7:56 am

    Well, well, I just found a bug in /usr/sbin/firewall_install.sh. See this line:

    REQUIRED_MODULES=”ip_tables ip_conntrack ipt_state iptable_filter ip_conntrack_irc ip_conntrack_ftp”

    Then further down:

    for MODULE in \$REQUIRED_MODULES; do
    if (( \`modprobe -l | grep -c “\$MODULE”\` )); then
    modprobe \$MODULE > /dev/null 2>&1
    fi
    done

    Some module names have changed over different kernel versions, and ‘ip_conntrack_ftp’ is now ‘nf_conntrack_ftp’, however ‘modinfo’ shows that the old name is an alias. However, ‘modprobe -l’ does not find the old alias, hence ‘nf_conntrack_ftp’ never gets loaded.

    I’ll fix this. Note, the firewall wizard GUI script that we use in Puppy is a dead project, so it’s up to us to update it.

  5. zygo Says:
    December 21st, 2007 at 5:00 pm

    With no LAN the wizard (used to) do the same as this except for line 4 which I added for cups

    iptables -P INPUT DROP
    iptables -P FORWARD DROP
    iptables -P OUTPUT ACCEPT
    iptables -A INPUT -s 127.0.0.1 -d 127.0.0.1 -j ACCEPT
    iptables -A INPUT -m state –state ESTABLISHED,RELATED -j ACCEPT
    iptables -A OUTPUT -p icmp -m state –state INVALID -j DROP

    There’s no state module in 4a3 so as Barry says the alphas aren’t for connecting to the wild wild internet.

    The match module would allow blocking/allowing of specific Process IDs. I.E. a basic application firewall I think.

  6. kirk Says:
    December 21st, 2007 at 5:35 pm

    Tried to compile rutilt 0.16 in dingo. Got the same error as you did. Downloaded and compiled gcc 4.2.2. Now rutilt compiles fine and the error with glxgears is gone.

  7. kirk Says:
    December 21st, 2007 at 6:08 pm

    Also, I didn’t specify an architecture when building gcc, so It built with i686.

  8. BarryK Says:
    December 22nd, 2007 at 8:02 am

    zygo, I have rolled back to the 2.6.23.12 kernel and made sure that every firewall-related module is turned on. That will be in alpha4.

    kirk, ha, ha, that was going to be my next step! Did you do anything else special when configuring, or was it just straightforward ‘./configure’?

  9. kirk Says:
    December 22nd, 2007 at 1:26 pm

    Yes, just did ./configure and make. If I would have thought about it, I would have specified i486 for the build environment, just to rule that out. But sounds like a T2 problem. If gcc is messed up, I wonder if that could cause other problems, since gcc builds everything.

    Also, the glxgears error message is gone, but I still can’t get it to work with DRI. It’s very strange. With DRI enabled, all the logs look good and glxinfo/glxgears -info looks good. If you try a game like Quake 3, it runs without error and you can hear the game running but the screen is black. Glxgears screen is black too and reports normal looking FPS. If I disable DRI in xorg.conf then glxgears actually shows the gears going around. The FPS seems a bit high though. Alpha2 did the same thing, so I installed the fglrx driver package from ATI. That worked and games played, but slower than in 3.01. Might just be my Radeon mobility x300 and Xorg 7.3, but has worked with Xorg 7.0 and 7.2. I’m curious to know if someone has DRI working.

  10. pakt Says:
    December 22nd, 2007 at 7:20 pm

    kirk, I read somewhere that Xorg 7.3 is unusually buggy and that whatever distro I was reading about had gone back to Xorg 7.2.

    Could explain your experiences with DRI.

  11. tempestuous Says:
    December 22nd, 2007 at 10:51 pm

    Yes, a few people have reported less-than-perfect results with the 3D Unichrome (VIA) driver under Xorg 7.3.

  12. linmodemstudent Says:
    January 20th, 2008 at 6:40 am

    I was googling around and noticed your comment regarding the ESS v0.3 driver not compiling. I haven’t tried it in 2.6.23 yet - would you like me to debug this?

    Cheers,
    Jeff
    (ESS driver maintainer)

  13. BarryK Says:
    January 20th, 2008 at 11:23 am

    Jeff,
    Yes please! But, I’m moving up to the 2.6.24 kernel. I’m waiting for the final release and then will bring out Puppy “Dingo” 4.00beta.

  14. linmodemstudent Says:
    January 20th, 2008 at 4:14 pm

    OK, just installed Fedora 8 (2.6.23.9 kernel) and it compiles OK for me, actually. I get some warnings, but it produces insmod-able .ko files. Can you pass on your error messages? Email is OK too.

    Cheers,
    Jeff

  15. linmodemstudent Says:
    January 27th, 2008 at 12:38 pm

    Hi Barry,
    I’ve got the warnings cleaned up now and the whole thing tested through 2.6.23. If you’re interested in a revised version (or a patch), just email me.

    Regards,
    Jeff

  16. BarryK Says:
    January 27th, 2008 at 7:04 pm

    I have sent a copy of the following to Jeff by email:

    jeff,
    I have now moved to the 2.6.24 kernel, and get the same errors as before.
    You mentioned fixing warnings, but I got unresolved symbols. One thing I recall that was missing … “jiffies” …or something like that — does that mean anything?

    Anyway, I will have a go at compiling your latest.

    Oh, one problem I had with most of the modem drivers, they wouldn’t compile with the 2.6.24 kernel. Going from memory again, but “SA_SHIRQ” symbol was undfined. This is in ‘interrupt.h’ in the kernel source, or rather it was up until 2.6.23.x kernel. It is deprecated and finally removed. I just copied the symbol
    declaration from an older kernel, and the modem drivers all compiled …dunno if they work, but ‘depmod’ reports no missing symbols.

    Regards,
    Barry

Gparted, Inkscape recompiled with static gtkmm

December 19th, 2007

As the gtkmm libraries are enormous, as big as the gtk libraries that they are a layer over, I have recompiled Gparted and Inkscape linked statically against the gtkmm libraries. So, there are now Gparted and Inkscape PET packages that do not have the ‘gtkmm’, ‘glibmm’, ‘cairomm’ and ‘libgnomeanvasmm’ packages as dependencies.

Note, compiling these libraries statically adds almost 2MB onto the executables, for example ‘gparted’ compiled with gtkmm shared libs is 575KB and compiled statically is 2533KB. However, that overhead is still far less than having to include the gtkmm shared libs in Puppy.

The ’standard’ Puppy has Gparted, and as I’m always trimming fat wherever possible, I hate it whenever I have to compile Gparted and throw in that 2MB of useless fat. I wonder how difficult it would be for us to write our own GUI partition manager?

As was reported, the icons for Pfind and Zfind in the ‘Filesystem’ menu in alpha3 are missing. This happened when I upgraded the Pfind package — previously there were two icons, ‘mini-arrow-right.xpm’ and ‘mini-search.xpm’ in /usr/local/lib/X11/mini-icons, but these are not in the latest Pfind package. The .desktop files for these packages specifies ‘mini-search.xpm’. Anyway, I have now placed these icons in the base collection in /usr/local/lib/X11/mini-icons so they will always be there.


7 Responses to “Gparted, Inkscape recompiled with static gtkmm”

  1. raffy Says:
    December 19th, 2007 at 7:12 pm

    Gparted is for occasional use only, and is needed only in conventional large machines. These could be the options: (1) users of large machines may simply be given a tutorial on how to do it in their existing system (very likely Win$); (2) design a download-use-discard scheme for Gparted; or (3) put it in zdrv.

  2. raffy Says:
    December 19th, 2007 at 11:34 pm

    Here is an example of future solid state drive (SSD) storage:
    http://education.zdnet.com/?p=1436
    “..the drive is the size of a fingertip .. future generations will hold up to 16GB.”

  3. disciple Says:
    December 20th, 2007 at 3:01 pm

    zdrv is a great idea. It might also be a good time to sit down and figure out if there is anything else that would suit living there.

  4. disciple Says:
    December 20th, 2007 at 5:23 pm

    Barry, have you considered putting a shutdown dialogue like the one here http://murga-linux.com/puppy/viewtopic.php?p=161867#161867 in Puppy?
    I think it is something that is really helpful for a lot of users, but one they are unlikely to get if it isn’t there to start with (i.e. how would they know that my one or any other exists?)
    If you need persuading, have a read of my reasons (and other people posted a couple more - the laptop mousepad one is particularly important).
    I think other people have made similar dialogues, and maybe someone needs to make it look better, but are you open to the idea?

  5. matzmann Says:
    December 20th, 2007 at 5:57 pm

    Jee, what a coincidence!

    Just these days I’m trying hard to compile a static binary of gparted for inclusion with INSERT.
    I compiled the static glibmm and gtkmm libraries and pointed compilation of gparted to use those, but it’s still running into a

    /root/gtkmm-2.12.3/gtk/gtkmm/.libs/libgtkmm-2.4.a(main.o): In function `Gtk::Main::init_gtkmm_internals()’:
    /root/gtkmm-2.12.3/gtk/gtkmm/main.cc:448: undefined reference to `Glib::init()’

    followed by a few more undefined references in the final linking process.

    Can you please give me hints or a recipe on how you compiled?

    You can reach me at please att insert dott cd

    Best regards
    Matthias

  6. BarryK Says:
    December 20th, 2007 at 6:33 pm

    matthias,
    I made some notes on another computer, that I don’t have access to right now, but basically what I did was fairly simple…

    I compiled the gtkmm packages with both static and shared libraries.
    I ran the normal ‘configure’ for Gparted.
    Before running ‘make’, I removed the ‘.so’ symlinks for the various gtkmm libraries.
    The build stopped at the final link step, complaining of missing libraries. I selected the final link step, which was displayed in the terminal window, pasted it and edited it, replacing the gtkmm ‘.so’ entries with ‘.a’. I ran that and it worked.

    Note, I also compiled Inkscape with static gtkmm libs, and this worked without having to do any hacking.

  7. matzmann Says:
    December 21st, 2007 at 10:30 pm

    Hi Barry,

    you made my day!

    This worked perfectly.

    Best regards and thanks
    Matthias

Puppy 2.14R updated to v1.01

December 18th, 2007

For anyone coming in new to this, and wondering why I’m announcing version “2.14″ when Puppy has moved on to 3.01 and my experimental Dingo Puppy, well, the “R” means “Revisited”. For those who particularly like the Puppy series from 2.12 to 2.16.1, or who have found that their PC won’t boot with later puppies, this Revisited puppy will interest you. Essentially it is many later ideas and improvements backported to a popular older Puppy.

Pakt and Dougal have been working intensely since they released version 1.0, and now version 1.01 is out. Others have contributed, for example temptestuous has updated some drivers. See forum announcement:

http://murga-linux.com/puppy/viewtopic.php?t=24574


One Response to “Puppy 2.14R updated to v1.01”

  1. JustGreg Says:
    December 19th, 2007 at 8:30 am

    I have been using Puppy 2.14R 1.01 with my ASUS EeePc. It is works well, especially with the Temptestuous pet files for the wireless and ethernet networking. See the forum topic that Barry references. Pakt and Dougal had done a great job in producing this version of Puppy.

Xine-ui, Xfmedia, xine-plugin

December 18th, 2007

I have found a combo that works ok, at least for me on my laptop. That’s Xine-ui and xine-plugin.

Xine-ui
This is a Xlib based GUI for Xine, that is, it doesn’t use GTK. This means that it’s  user interface looks and behaves nothing like GTK user interfaces. It works though, which is what we are desperate for. Full-screen also works. My testing was fairly limited, but JWM behaved itself and rendered both the control-panel and video windows nicely on-screen.

Xine-plugin
This plugin for Mozilla browsers has nothing to do with the Gxine plugin. It is a plugin that plays videos inside the browser window, not in an external window — for example the Gxine plugin launches Gxine to play videos. It may have some disadvantages, as it lacks the controls that the external player has. Anyway, I did some basic testing and it works.

Xfmedia
XFCE packages are a problem to compile in Dingo, as they can’t be. There is a bug during compiling where glibc tries to call a function that does not exist. I googled around and there is very little information but the reports mostly seem to be for glibc version 2.6.1 on x86, same as Dingo uses. However, XFCE does compile in T2, so that’s what I used.
Xfmedia needs libexo, libxfce4util and libxfcegui4 packages, and I compiled that and tested in Dingo. Xfmedia is rather limited and buggy. When I try to open and play an audio file, it does something really weird and the file cannot be played. When I tried to play a DVD video, the video-window was off-screen (JWM fault) but also was only several pixels wide and high and the only way to get a video window to display correctly was to specify the size on the commandline. Full-screen does not work. Too many problems.

Note, given the problem with compiling XFCE, I plan to compile all of the XFCE packages in T2 and make them available as PET packages, for the convenience of those wanting to build a custom puppy using XFCE.


2 Responses to “Xine-ui, Xfmedia, xine-plugin”

  1. lobster Says:
    December 18th, 2007 at 7:04 pm

    “XFCE, I plan to compile all of the XFCE packages in T2 and make them available as PET packages, for the convenience of those wanting to build a custom puppy using XFCE”

    Very good news. XFCE is good looking and stable as is Icewm. JWM is stable too and has been for a while. A couple of things I liked very much in Dingo was the ability to change the icon set and the JWM theme

  2. plinej Says:
    December 18th, 2007 at 9:49 pm

    Yeah, I’ve noticed that compiling bug with other things as well. Fluxbox is another example, there are others too.

Gparted, firewall bugfixes

December 17th, 2007

These are tentative bugfix reports.

In Dingo alpha3, Gparted crashes when any operation is performed on a partition. The Debian people also have this problem:

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=446522

Operations do complete, but the crash occurs when the partitions are being rescanned. They have posted a patch:

http://bugzilla.gnome.org/show_bug.cgi?id=470387

It seems that bugfixing on the Gparted project is moving very slowly right now, and this patch has not been applied.

I am pre-announcing this bugfix, hope it will fix our crash problem also.

Regarding the firewall, it was reported to be broken. I have just compiled the 2.6.23.11 kernel, and took extra care to make sure all firewall-related modules were turned on, and this seems to have fixed it. That is, in alpha3 there was probably one or two modules missing.

Note, I compiled the 2.6.23.11 kernel to find out if that would fix Gparted, but it still has the same crash.


7 Responses to “Gparted, firewall bugfixes”

  1. Sage Says:
    December 17th, 2007 at 4:38 pm

    I cannot test 4alpha3 for all the reasons stated, but wonder if the bizarre failure of GPartEd requiring (but not using) insertion of a FD before it completes start up is addressed. This also occurs sometimes in the 3series.

  2. BarryK Says:
    December 17th, 2007 at 6:22 pm

    Yippee! The patch for Gparted worked, at least it didn’t crash after I changed a partition flag.

    Sage, the floppy disk thing is a bug in ‘parted’, one of the packages that Gparted uses. I think that bug is known to them for sometime.

    Sage, when you report “cfdisk does not work” you do not give any details. You have reported this many times. Whenever I test cfdisk, it works fine. Is there any specific operation that fails?

  3. Sage Says:
    December 17th, 2007 at 8:22 pm

    Sorry about the cfdisk thing. It’s been an on-off thing with recent releases; this is an OFF version. It’s simple: type at the console prompt and either the partition configuration pops up or a textual error message on the bottom line. It’s usually the same error message that appears when there’s an operator error, so I always check with an earlier Puppy version that I know is reliable. Obviously, I wouldn’t report artefactual problems or finger trouble!

  4. Sage Says:
    December 17th, 2007 at 8:24 pm

    Huh! the cfdisk between ‘more than/less than’ brackets seems to be omitted on this blog. Please insert between ‘type’…..’at the console..’

  5. Sage Says:
    December 17th, 2007 at 8:58 pm

    The FD feature is also irregular - I’ve only had it on the 3 & 4 series.

  6. raffy Says:
    December 18th, 2007 at 8:02 am

    Test - inserting more than/less than brackets:
    - the usual, which may not show:
    - the use of amperand+lt/gt+semicolon: <this>

  7. zygo Says:
    December 19th, 2007 at 7:16 am

    I hope this includes the match firewall module. It allows blocking/allowing of specific Process IDs. I.E. a basic application firewall.

    Will the next release be this weekend?

devx_393.sfs re-uploaded

December 14th, 2007

I just uploaded ‘devx_393.sfs’ to ibiblio again, as I found a bug in the previous file.

I have compiled the NetSurf web browser from SVN. This is a really tiny browser that can do the job for the planned ‘barebones’ Puppy. It supports CSS, so can be used for the CUPS web interface, and also as the internal HTML help viewer. Plus, it is good enough for web browsing on sites that do not require a Javascript interpreter to render pages and provide user interaction.

NetSurf home:
http://www.netsurf-browser.org/


One Response to “devx_393.sfs re-uploaded”

  1. raffy Says:
    December 16th, 2007 at 12:00 pm

    Sorry for the off-topic, Barry, but I guess we need assistance from servage.net to update the “bad-behavior” plugin placed there by the previous admin. This update is needed:
    http://www.bad-behavior.ioerror.us/2007/12/06/bad-behavior-2011/

    “..users have found themselves blocked from their own sites while using recent versions of Bad Behavior. A third party blacklist which Bad Behavior queries recently began sending false positives for any IP address queried, causing everyone using Bad Behavior to be blocked. This issue is fixed in Bad Behavior 2.0.11.”

    I can’t do admin tasks while this error is happening, and ssh access may be needed to find and update the bad-behavior files.

    puppylinux.org/user may be taken down but perhaps not the wiki. Either way, admin access is needed to do maintenance work, which is not possible presently.

(c) Copyright Barry Kauler 2008, all rights reserved. http://puppylinux.com/