Get Wary from here:
Two little bugs
After I had just started to upload the ISO, I discovered two buglets. Guys, the first time that you boot this pup, would you mind immediately open a terminal window and type these two lines:
# echo -n "1" > /var/local/sns/rx_bytes_session
# ln -s depmod /sbin/depmod-FULL
The reason for the first line is that the Video Upgrade Wizard checks /var/local/sns/rx_bytes_session to see if there is an Internet connection, however that file doesn't even get created until the first shutdown. It is only a buglet, it just causes a message in the Video Upgrade Wizard telling you that you don't have an Internet connection even when you do.
The second buglet is caused because I have gone back to the full 'depmod', but somehow neglected to create 'depmod-FULL' as a symlink to 'depmod'. The name 'depmod-FULL' is needed by some scripts.
I will fix those two bugs before I upload Woof.
Video Upgrade Wizard
You can run this via clicking on the desktop 'setup' icon, or in the 'Setup' menu.
This Wizard is supposed to analyse your current X situation and if appropriate offer a driver upgrade. I have tested the Intel video, but not the latest nVidia -- for the latter, if the Wizard sees that you are only running the pup with the 'vesa', 'Xvesa' or 'nv' drivers, then you will have an offer to install a commercial nVidia driver -- I have created three nVidia PETs for the 188.8.131.52 kernel and the Wizard is supposed to detect which one of those will work on your video hardware and offer that.
Anyway, let me know if anything does not go according to plan. If you do install the commercial nVidia PET that the Wizard recommends but it doesn't work, let me know also.
Note, the Intel upgrade driver has to run the kernel with KMS enabled, which requires a reboot after installing the PET. There are popup instructions that explain that, they also state that you need to reboot with "pfix=nox" and then run "xorgwizard", however the "pfix=nox" is probably not required as X will fail to start anyway and will just drop you to the console. Note that with the KMS-enabled kernel none of the other X drivers (vesa, Xvesa, i810) will work.
Analog modem drivers
For those who have dialup analog modem cards, the Wary build with 184.108.40.206 kernel has a good collection of drivers, including for the Agere chips. However, drivers for Intel, Conexant and ESS modem chips are missing, and for these you will need the Wary build with 220.127.116.11 kernel.
Wary 0.9 uses the 18.104.22.168 kernel, but I have rolled back due to show-stopper bugs in that kernel.
There is also a show-stopper bug in the 22.214.171.124 kernel, so I am using an especially patched 126.96.36.199. Test this guy, let me know if this is "the one" for Wary.
Welcome 1st boot
The first time that you boot Wary (or with pfix=ram) you get a little message at the top of the screen inviting you to mouse-over for getting-started information. This will cause an information window to popup -- this has been created by gtkdialog, but there may be an issue with the dimensions. If the information window is too high/wide or too short/narrow, let me know, also what is your screen resolution (ex: 1280x800).
This problem occurs because gtkdialog has a bug, cannot auto-size it's window properly when the window has text with html markup, so I call gtkdialog with the --geometry option (see script /usr/share/doc/welcome1stboot).
Upgrading an existing warysave.2fs
When you boot Wary for the first time or with "pfix=ram" from CD, at shutdown you are asked if you want to copy wary_092.sfs off the CD to HD (and zw092346.sfs, the "zdrv"). With earlier puppies, it used to get copied to HD anyway at bootup, but users objected to that, so it no longer happens.
The outcome of this is that if you boot Wary 0.9.2 and there is an existing warysave.2fs personal file, it will get upgraded and you will get to a desktop -- and the wary_092.sfs and zw092346.sfs will not be copied to HD.
This slows bootup, as they have to be read from the CD at bootup, rather than from HD.
I need to think about this. I will probably need to modify the 'init' script, detect that a warysave.2fs file is about to be upgraded, see if the wary_092.sfs is not on HD and offer to copy it from CD. I have added this to the to-do list.
For now, if you want you can copy the .sfs files to HD manually.
Wary 5.0 release schedule
From now until the release of Wary 5.0-final, I want to focus on bug-fixing of Wary and Woof and hold-off any significant changes. So, although we always have ideas for improving Wary and Woof, if it isn't essential then I want to leave it for now.
I can't really say how far away 5.0-final is. It depends on feedback on 0.9.2. If there are only little buglets then great, I can maybe bring out an RC in 4-5 days ...but, I had better not make premature statements, as I have just moved to a new kernel and lots of bugs may emerge...
Comments:Posted on 24 Oct 2010, 13:56 by BarryK
Wary 092 forum feedback
I have started a forum thread:
Posted on 24 Oct 2010, 13:58 by BarryK
Mirror for Wary 092
Here is a faster mirror for downloading Wary:
Posted on 24 Oct 2010, 14:34 by Sage
...doesn't work for me at present, just displays pageful of code. I've seen this before, rarely, eg with Lighthouse. ibilio is good and fast this morning, though.
Posted on 24 Oct 2010, 15:25 by Terryphi
Welcome 1st boot popup
Welcome 1st boot info popup looks fine to me at 1024x768 resolution.
Posted on 24 Oct 2010, 15:41 by Sage
Welcome 1st boot popup
Sorry - not quite @800x600 on a 17"lcd, onboard SiS754 board. The Welcome panel cannot be moved and the lower portion is obscured.
Posted on 24 Oct 2010, 15:51 by BarryK
What is Wary?
Some of you guys who are newcomers or just read my blog sometimes, might not be familiar with what Wary Puppy is all about. To get the background, read this page:
Note though, although Wary is targeting older hardware, we have X upgrade packages (see the Video Upgrade Wizard in Wary 0.9.2) that will get very recent video hardware working -- for example with the Intel upgrade my Acer netbook with Pineview video works great (prior to installing that, the generic 'vesa' and 'Xvesa' drivers worked but they are slow).
Posted on 24 Oct 2010, 15:59 by BarryK
For anyone who needs to compile a driver that needs the kernel source, the 'kernel_src-188.8.131.52-patched.sfs' file is here:
Us#er#na#me: pu#pp#y Pa#ss#wo#rd: li#nu#x
Just a reminder, in Wary 0.9.2 'depmod' is now the full version. Also, the kernel source sfs file already has 'make prepare' executed prior to its creation.
Posted on 24 Oct 2010, 16:01 by Sage
"...what Wary Puppy is all about"
Is this aimed at YT? Although it doesn't seem to fit. The only machine I had available for a rapid feedback, as requested, was a 754. I will try to dig out something really old later in the week, but I left most (all?) of my 486 s & PI/PII s behind after moving. Maybe someone else will take up the challenge.
Posted on 24 Oct 2010, 16:14 by BarryK
Re old computers
The Wary intro page explains, anything about 2 years old or more.
Back to the earliest Pentium/686. I am not going to support 486 CPUs. Mplayer and ffmpeg require a 686.
"Cannot be moved" -- actually it can. Standard technique, hold down the ALT key and use the mouse to drag. Hm, I might have to tell the window to draw from top-left of screen.
# ln -s depmod /sbin/depmod-FULL
does work. You must have typed it wrong.
Posted on 24 Oct 2010, 16:49 by peebee
Agere HDA old version
The Agere HDA modem in K2.6.34 is the old 2.1.60 version from 2005 rather than the 2.1.80 version that rerwin and I have been testing recently. As a result it crashes on my HP 550 laptop.
I haven't tested 2.6.30 yet....
Posted on 24 Oct 2010, 17:04 by Sage
"...what Wary Puppy is all about"
How time rolls on! 754 is rather more than 2yrs old, already! [But I don't believe in multi-cores anyway; they don't run any faster and there's an additional overhead whilst tasks are farmed out to the correct core and subsequently re-correlated. Smoke & mirrors.]
"...moved" - ah, yes. Done that but the lower part is still clipped on the SiS Mirage 1.
Re. '486' , quite a few folks report good results with the KII/KIII-500/550, (which can be clocked, of course) and were the pre-cursors for true Athlons, which were introduced at 650(?). [What has changed in the basic x86 architecture since 286?! ARM sells more and TI-68000/Apple blew it back in the 80's when they got so greedy.]
"# ln -s depmod /sbin/depmod-FULL" - sometimes difficult to discern spaces, that's my excuse; non-coders wouldn't be sure in some fonts if they aren't familiar with statement structures. Any road up, I seem to have typed the correct version with the scsi-version now loaded!
Posted on 24 Oct 2010, 17:40 by Terryphi
Bad news: OpenOffice sfs crashes Wary 092 for me.
It happens after the registration screen when the OO splash screen appears.
I am testing on a clean frugal install.
If it is of any relevance, yes I did the 2 fixes required on first boot.
Posted on 24 Oct 2010, 19:00 by BarryK
Hmmm, I don't know how I still have the older source. I think you already did this, but would you mind again posting links to the latest source?
Posted on 24 Oct 2010, 19:03 by BarryK
That was one of the things that I checked, it worked for me. That was on the Acer netbook, with OO 3.1.1.
I'll test again.
Posted on 24 Oct 2010, 19:25 by BarryK
Oh crap, it crashes on my main laptop.
What video driver is being used on your computer?
I wonder if that is a common factor. This laptop uses the 'intel' Xorg driver.
I'll do some more testing.
Posted on 24 Oct 2010, 21:05 by Terryphi
OO and Xorg
Barry: I am using the intel Xorg driver too.
Beyond the OO issue, Xorg is still troublesome for me. If I set xorgwizard for 1024x768 resolution it is not saved at next boot. There is definitely a problem with Intel 82G31/33. I am about to try a version of xorg.conf that worked reliably in Wary 070.
Posted on 24 Oct 2010, 21:40 by Terryphi
Xorg G31/G33 - 1024x768
No, Wary 092 will not reliably save a working xorg.conf set to 1024x768 on next boot with this chipset.
Here are the last few messages on reboot:
Intel 800/900 Series VBIOS Hack version 0.5.3
BIOS: TYPE 1
Mode Table Offset: $C0000 + $269
Mode Table Entries: 27
Patch mode 38 to resolution 1024x600 complete
Status X, ......
If X failed to start, type "xorgwizard" to setup X
Posted on 24 Oct 2010, 22:49 by Terryphi
Xorg Intel G31/G33 -1024x768
Xorgwizard seems unable to generate a suitable xorg.conf for this chipset which works on reboot. The only solution is to manually amend xorg.conf, restart Xserver and reboot. Some possible amendments work when Xserver is restarted but seem to be rejected by Xorg as invalid on reboot.
These amended sections work for me
VendorName "Monitor Vendor"
ModelName "Monitor Model"
#UseModes "Modes0" #monitor0usemodes
Option "PreferredMode" "1024x768"
HorizSync changed to 40-80 and VertRefresh changed to 60.
#Option "metamodes" "1024x768_60 +0+0" #METAMODES_0
Depth changed to 24.
Posted on 25 Oct 2010, 14:42 by 01micko
gtkdialog-splash exit on mouse-over
Just for some things the mouse-over exit for gtkdialog-splash is inappropriate.
For example.. I am just doing a frugal install using 'universal installer' of wary-092 and if I accidentally mouse-over the splash it disappears and I don't know what's going on... this could be an issue on older slower machines. Maybe the 'never' option would be better in this instance (and possibly some others)
Posted on 25 Oct 2010, 16:29 by peebee
Some more Agere HDA feedback....
I've asked rerwin by pm to let you know where he got the 2.1.80 sources from.....
I've now tested K2.6.30 and its different to 2.6.34 - the Agere modules are there but don't get loaded at all - unfortunately I can't get any diagnostics for you as dmesg, /var/log/messages and modprobe in a terminal are all silent as to why the module doesn't load....
BTW - wary060 K2.6.32 had the 2.1.80 modules.... did you have the sources then?
Please say if there is anything else I can do to help.
Posted on 25 Oct 2010, 16:37 by peebee
"In fact, I one of the goals for Wary is that media playing will be "the best", handling just about every media format."
I think?? the version of mplayer in wary092 lacks digital TV support? At least I can't find it! The version in luci232 does provide this.
Posted on 25 Oct 2010, 16:46 by BarryK
This is where I got the Agere sources from:
I am using:
The first one is version 2.1.60 but that appears to be the latest available.
If you know of later source for it, please let me know.
Posted on 25 Oct 2010, 20:13 by peebee
How odd - your sources look correct to me - rerwin can maybe confirm definitely - but the modules in wary092 k2.6.34 are identifying themselves as the older 2.1.60 for the 11c11040 which is what I have - and crashing. You can confirm for yourself I think by modprobe agrmodem and then look in pupscan interface info.
Posted on 25 Oct 2010, 20:27 by peebee
Agere - more diagnostics
/lib/modules/184.108.40.206/agrmodem has 3 subdirectories - 06pci, 11c11040 and 048pci and a symlink called active which in my case is correctly linked to 11c11040 however pupscan shows that it is the 048pci module that has been loaded which is why version 2.1.60 is showing I guess. So the wrong Agere type is being loaded.
Sorry I didn't find this earlier.
Posted on 25 Oct 2010, 23:03 by Terryphi
OpenOffice crash -solved for me!
I just added java-jre-6u20.pet, restarted Xserver and OO no longer crashes!
DaveS and tronkel do not confirm this as a solution but they may be using a different version of Java or complicated their systems with LibreOffice as well as OpenOffice.
Posted on 25 Oct 2010, 24:40 by Terryphi
OO works on kernel 220.127.116.11 too!
Elated with my success with OO on Wary 092 I have revisited a clean install of Wary 090 kernel 18.104.22.168. Guess what? Add java-jre-6u20.pet before trying the OO sfs and then it does not crash. Yippee!
How did we all miss the importance of Java?
Posted on 26 Oct 2010, 4:46 by BarryK
Re OO good news
That is good news. I wonder, perhaps we should re-build the OO SFS with the java inside it.
Posted on 26 Oct 2010, 14:13 by Terryphi
Barry: I would prefer not to see Java included in a OO sfs. Java is not needed for most OO functions and is an ongoing security liability. When I have it on a system I switch it off in OO and all browsers.
Apparently some distributions have built OO packages which have disabled parts of OpenOffice.org which rely on Java. I read somewhere that the Fedora rpms have done this. I don't know about the Ubuntu debs which are based on Go-OO.
I think that the LibreOffice developers are coming under pressure from users to ditch the Java dependency. Given the politics of their split with Oracle hopefully that might happen!
Posted on 26 Oct 2010, 17:50 by Terryphi
Even more on OO sfs
I hope this is not getting boring. I've done some more testing with OO sfs and the 22.214.171.124 kernel, although I assume it has wider relevance. Once OO is installed it does not need Java to run. I had switched off Java in OO's options. I have successfully tested swriter and scalc which are the only ones I use.
After the java-jre-6u20.pet had been removed these files are found in Wary which may (or may not!) be relevant:
Anyway, I'm thinking that possibly very little needs to be added to the OO sfs to keep OO happy the first time it is loaded.
Posted on 27 Oct 2010, 24:19 by Terryphi
OO crash fix
I have now established that only one file, and the sub-directories to locate it, are all that is necessary to solve the OO crash on kernels 2.6.34 and 126.96.36.199 .
Details on the forum:
Posted on 28 Oct 2010, 23:06 by Terryphi
LibreOffice and go-OO need Java
I have extended my testing to the go-OO and LibreOffice sfs files. My fix for the OpenOffice crash does not work for go-OO and LO. They are much more heavily reliant on Java. That is a very good reason to stick with OpenOffice 3.1.1.