Precise Puppy RC2, October 20, 2012

[oldyeller]

Hello everyone,

I have done a woof2 build here Precise 5.2.60.

I am new more or less to woof2, I am not sure were this build is to BarryK's
Hope you all will enjoy this continuation of precise

Cheers

[BarryK]

Dang it, just when Puppy was starting to move ahead of the big dogs someone tossed a little raspberry pi in front of it and brought things to a screeching halt. Been pretty quiet around here since then unfortunately.

I'm baaack!

Precise Puppy beta3 is coming soon, with 3.2.27 kernel and latest Woof goodies.

This evening I plan to start reading this forum thread, from page 15 which was when beta2 got released.

Yes, as 01micko mentioned, I am planning on writing a "big app" and slowing down the pace of Woof/Puppy development.

Well, I do make plans... but it doesn't always turn out that way. Those who have known me for a long time know how many times I have planned to "retire".

Anyway, Precise beta3 is on the way!

[cthisbear]

BarryK:

Yes! in our books retirement suits you...ha! ha!.

Hope you are enjoying Perth BK.

Regards.....Chris.

[don570]

Note to BarryK:

Have you looked at how Shinobar organises his Japanese distribution
into separate SFS packages? It has potential.

The Porteus distribution uses something similar but calls them
'XZM Modules' You store these modules in a specified folder
and they load automatically during boot process. ...But there
is an option to not load modules or change the location of
modules folder.

I was able to get his Japanese Seamonkey SFS to work in Precise.

The goffice contains Abiword ,gnumeric and Inkscapelite
but I couldn't get it to work in precise.



_____________________________________________

[don570]

I got an idea while looking a Tazoc's Lighthouse 503

He puts a mhwaveedit configuration file in his ISO
so that the first time a user launches mhwaveedit
it will have a wide window and less vertical scroll bars.
It's a nice touch!!

Here's the file he uses ----> /root/.mhwaveedit/config
Note that there is a line for window geometry

Code: Select all

# mhWaveEdit configuration file.
# Automatically generated by mhWaveEdit 1.4.20
# May be hand edited but extra comments will be removed when the 
# settings are saved.
# Remove this file to restore default settings.

tempfilesFP = true
speedReset = false
viewQuality = 128
timeDisplayScale = 0
varispeedConv = 5
tempDir1 = /root/.mhwaveedit
rollCursor = false
mixerUtility = retrovol
showTimeScale = true
showSpeed = false
showHorizontalZoom = false
playerFallback_BigEndian = false
showVerticalZoom = false
speedSmooth = true
speedConv = 0
soundDriver = auto
soundBufferSize = 65536
timeDisplay = 0
mainwinFront = true
recentFiles = 4
drawImprove = true
varispeed = true
playerFallback_SampleSize = 2
autoPlayMark = false
outputStereo = false
useGeometry = true
outputByteswap = false
ditherPlayback = 0
windowGeometry = 50_50_822_292
diskEditingThreshold = 131072
centerCursor = true
soundLock = false
playerFallback_Signed = true
ditherEditing = 1

__________________

[cthisbear]

" don570 ...Note to BarryK:

Have you looked at how Shinobar organises his Japanese distribution
into separate SFS packages? It has potential. "

::::::::::

Lupu-528JP (Japanese/English bilingual) is released.

An excellent piece of work from Shinobar.

Chris.

[BarryK]

Gtkhash is still not working so I downloaded the source and made a pet of it.

Thanks, I have uploaded the PET to ibiblio:
http://bkhome.org/blog/?viewDetailed=02954

[BarryK]

Geany has "funny" colors in screens with 16 bit depth.
I remember having seen some comment on Pemasu's Precise but no solution.
Even if you get new color schemes for geany, the problem persists.
Googling for it, looks like an Ubuntu problem though I came across a likely bug in Fedora 17.
Anyone has any idea?

I do not have any idea but I compiled Geany from the latest git in Precise 5.2.60 using the defaults and works fine.
No green background color.
The exe, dev, doc,and nls pets are all together here, in case someone with this problem needs them (md5sum: 5a1f194888d1121dcaa6c0adb650bc32)

Yeah, that coloured background is weird.

For beta2 I am using Geany PET from Wary.

[BarryK]

Thanks Barry for making a link /usr/lib/gtk-2.0/2.10.0/gtk.immodules.
But still we have a problem.

The SCIM in lang_pack_ja worked on Pemasu's precise and on Barry's alpha-3 by the treatment above.
But on the beta-1(EDIT: and beta-2), SCIM launches and can input KANA letters but fails to convert int KANJI letters. Pressing the space-bar to swithing the mode but does not work.

We are aware that the /usr/lib/gconv is cut down. So we add japanese stuff with the lang_pack_ja. Of cource we need to refresh the cache running 'iconvconfig'.

The problem is the file /usr/lib/gconv/gconv-modules is also cut down. We reverted the file with the one of the alpha3 and re-run 'iconvconfig'. Now the SCIM works OK,

There is a room to condider the file gconv-modules to be cut down or not...

Yeah, I changed the 'glibc' template in Woof a little while ago, so the full-size gconv-modules will be used.

[LateAdopter]

Hello BarryK

In the DOTconfig of the 3.2.27 kernel that you have just compiled it has this:

Code: Select all

CONFIG_DRM=m
CONFIG_DRM_KMS_HELPER=m
CONFIG_DRM_TTM=m
CONFIG_DRM_TDFX=m
CONFIG_DRM_R128=m
CONFIG_DRM_RADEON=m
# CONFIG_DRM_RADEON_KMS is not set

The Xorg instructions for Radeon kernel modesetting are here:

http://www.x.org/wiki/radeonBuildHowTo# ... odeSetting

It says there should be:

Code: Select all

CONFIG_DRM_RADEON=m
CONFIG_DRM_RADEON_KMS=y
CONFIG_FRAMEBUFFER_CONSOLE=y
# CONFIG_FB_RADEON is not set

Do you think this could be why the current Radion GPU driver stack does not work on Precise Puppy? KMS is a requirement for the R600g driver.

The other Puppy devs say that they use your DOTconfig. So, if this is wrong, their kernel will not work with the current Radeon drivers, either.

Thanks

[BarryK]

- missing libraries in /usr/lib/ :
ldd /usr/lib/libbonoboui-2.so
libcanberra.so.0 => not found
ldd /usr/lib/libedit.so
libbsd.so.0 => not found
ldd /usr/lib/libgnome-2.so
libcanberra.so.0 => not found
ldd /usr/lib/libgnomeui-2.so
libcanberra.so.0 => not found
ldd /usr/lib/libIntelXvMC.so
libxcb-dri2.so.0 => not found
ldd /usr/lib/libva-glx.so
libva-x11.so.1 => not found

Thanks for that, missing pkgs added.

[BarryK]

- missing libraries in /usr/bin/ :
ldd /usr/bin/cddb_query
/usr/bin/cddb_query: /usr/lib/libcdio.so.12: version `CDIO_12' not found (required by /usr/bin/cddb_query)
ldd /usr/bin/gnome-open
libcanberra.so.0 => not found
ldd /usr/bin/gphoto2
libcdk.so.5 => not found
ldd /usr/bin/gtkhash
libgtk-3.so.0 => not found
libgdk-3.so.0 => not found
ldd /usr/bin/jiv
libglut.so.3 => not found
ldd /usr/bin/sftp
libbsd.so.0 => not found
ldd /usr/bin/spumux
libMagickCore.so.4 => not found

Thanks, fixed.

[BarryK]

The problem is the file /usr/lib/gconv/gconv-modules is also cut down. We reverted the file with the one of the alpha3 and re-run 'iconvconfig'. Now the SCIM works OK,

Got a side-effect with abiword, it got unstable. I suspect the gconv-modules does not match with the modules contained, because we did not add fullset, so the iconvconfig makes imcompleted cache.

Maybe solved if we rebuild the gconv-modules with the modules currently containd (maybe we can find the code in woof, right?), but we take a tentative solution for the japanese:

1. Add gconv modules for Japanese.
2. Replace the /usr/lib/gconv/gconv-modules with the one of Ubuntu fullset.
3. Remove /usr/lib/gconv/gconv-modules.cache.

My experience has been that the 'iconvconfig' utility creates a broken cache file. I went down that route before, had to go back to either the full cache file as supplied by the original libc package, or no cache file.

It would be a very painful choice to go for the full set of files in /usr/lib/gconv, given Puppy's history of keeping things as small as possible. Our cutdown collection occupies 669KB, the full set is 6452KB (compresses to 2.5MB) in Ubuntu Precise.

But, it seems that the cutdown choice of files, plus the original full cache file, works. So I will keep that arrangement.

[pemasu]

To compile vlc with subtitle support, you need more files to the /usr/lib/gconv. I noticed that when I included the whole gconv folder content. Vlc compile with subtitle support succeeded.
The result was confirmed by Billtoo also.

http://www.murga-linux.com/puppy/viewto ... 933#645933

That thread has several other Precise Puppy compiled apps more.

[BarryK]

Precise Puppy 5.2.60 (beta2) frugal
Racy530 frugal

sylpheed-3.1.4
sylpheed-3.2.0beta7
sylpheed-3.2.0beta8

All FAIL to compile on precise5260
All compile OK on racy530

When compiled on racy530, all will run on precise5260!

I have compiled Sylpheed 3.2 in Precise Puppy.

I have created a PET '0hack1_precise_dev' that fixes the devx so Sylpheed will compile.

It is a difficult situation, as Debian/Ubuntu do not put the /usr/lib/*.la files from their DEBs. I think this is because of their multi-architecture dementia.

Anyway, I have created some .la files in my "hack" PET.

I will upload the Sylpheed PET soon.

[oldyeller]

I am very sorry for the confussion of Precise builds that I have done.
I have reposted with a notice that they are not official and that Barryk's Precise is.

[don570]

Advanced Bash Scripting guide updated 2012 is available


http://www.murga-linux.com/puppy/viewto ... 151#648151

_______________________________________________

[BarryK]

Hello BarryK

In the DOTconfig of the 3.2.27 kernel that you have just compiled it has this:

Code: Select all

CONFIG_DRM=m
CONFIG_DRM_KMS_HELPER=m
CONFIG_DRM_TTM=m
CONFIG_DRM_TDFX=m
CONFIG_DRM_R128=m
CONFIG_DRM_RADEON=m
# CONFIG_DRM_RADEON_KMS is not set

The Xorg instructions for Radeon kernel modesetting are here:

http://www.x.org/wiki/radeonBuildHowTo# ... odeSetting

It says there should be:

Code: Select all

CONFIG_DRM_RADEON=m
CONFIG_DRM_RADEON_KMS=y
CONFIG_FRAMEBUFFER_CONSOLE=y
# CONFIG_FB_RADEON is not set

Do you think this could be why the current Radion GPU driver stack does not work on Precise Puppy? KMS is a requirement for the R600g driver.

The other Puppy devs say that they use your DOTconfig. So, if this is wrong, their kernel will not work with the current Radeon drivers, either.

Thanks

I am compiling 3.2.28 right now, with Radeon KMS enabled.

I also went through the digital radio/tv section, enabled some more drivers, as pemasu mentioned someone requested one of these.

[rerwin]

Barry,
I have posted to oldyeller's precise thead here:
http://www.murga-linux.com/puppy/viewto ... 379#647379
the newest version of usb_modeswitch, as pets ready for woof-based puppies.They contain the components as generated by the source tarballs. Their pinstall scripts make your modifications during installation, so can be migrated to new versions as they appear.

I also posted the change log with the version I made for lupu, in its thread:
http://www.murga-linux.com/puppy/viewto ... 351#646351

Please consider adding the packages to the beta3. Thanks.
Richard

[rerwin]

Barry,
Thank you for adding my "fixes" to woof -- and for catching these misguided mods:

• rc.shutdown: #120603 rerwin: Retain deletions in first pupsave file by copying whiteout files.
  pup_event_backend_modprobe: #120608 . . .; for rule-loaded modules either blacklisted or supported by modaliases, load no module.

They confirm that I need a code inspector.

For firmware.dep, the deletions were unintended. I assumed I was modifying a pure-woof version of the file, but apparently not.

The bcma association was added in support of peebee's work with the Broadcom drivers, which involves several complex modprobe.conf files, including one for bcma.ko. Adding the association to woof is part of my effort to minimize the need for developers to change puppy code. But it may be unnecessary if my experiment with moving most of the complex .conf logic into backend_modprobe pans out.
Richard