January Extensions


Forum: The Testing Area
Topic: January Extensions
started by: roberts

Posted by roberts on Jan. 04 2008,15:48
Thanks goes to JasonW for

madbomber-0.2.5.dsl
madbomber-0.2.5.unc


Read the info file for requirements.

Now posted in the repository.

Posted by Juanito on Jan. 09 2008,03:38
Hi, I just noticed that the description for the bt_modules-2.4.26.dsl extension in the modules section is incorrect (my mistake), could you replace "bluetooth modules for dsl 4.x" with "bluetooth modules for dsl 3.x" in the info file please?
Posted by roberts on Jan. 09 2008,15:59
Done.
Posted by roberts on Jan. 13 2008,06:35
New in the repository....

Thanks goes to WDef for an update to:
gnupg2.uci

Thanks to JasonW for:
konqe.dsl   <- new browser, check it out!

Thanks to mikshaw for an update to:
lighttpd.uci

Also note that I am also now updating a < Master List > of all extensions.
Use your browser to search it, or download it and use Beaver to search it. Or have fun with grep.

Posted by roberts on Jan. 13 2008,20:22
Thanks goes to JasonW for

gnubg.dsl    GNU Backgammon.

Read the info file for requirements.

Now posted in the repository.

Posted by roberts on Jan. 15 2008,20:31
Thanks goes to castillo for:
tcpdump-3.6.dsl
nemesis-1.3.2.dsl

Posted by roberts on Jan. 16 2008,00:29
Thanks to Lucky13 for:
slock.dsl

Posted by lucky13 on Jan. 16 2008,00:51
Quote
Thanks to Lucky13 for: slock.dsl

Please don't use it. Robert, please remove until I can fix it. Sorry. :-\

EDIT... I've fixed and resubmitted, but please wait for Robert to post it before attempting to use it. The one that works has the following md5sum (changed for reason below):
278b83678762c57bad895910fb3e8271  slock.dsl

I also removed the menu entry because I don't want someone to invoke it unintentionally without setting a password. You must set a password for your user (dsl). There are no prompts. You type the password. Success, you get back to your desktop. Failure, console beep. It isn't fancy, but it's only a few kb (and unlike xscreensaver, there's no daemon or CPU wasting graphics to mess with).

Posted by roberts on Jan. 16 2008,16:14
OK. Removed.
Posted by curaga on Jan. 20 2008,14:02
Two extensions coming:

< Gnucash.tar.gz - gtk1 >
< Popt_dev.dsl >

All three files included in those (ext, md5, info)
(Gnucash was too big for my email, and popt-dev just came along)

edit: gnucash pack is not packed with bz2, seems I mistakedly used -z instead of -j with tar.

Posted by roberts on Jan. 20 2008,17:45
Thanks to Lucky13 for:
slock.dsl
screen.uci


Thanks to Juanito for:
svn.uci
cairo-1.uci
gimp-2.uci
xorg72.uci


More to come as soon...

Posted by roberts on Jan. 20 2008,18:16
Thanks to Juanito for:
samba-3.uci

Posted by Juanito on Jan. 21 2008,11:36
...and thanks to Mikshaw for pointing out the error in svn.uci and Meo for alpha testing cairo-1.2.uci and gimp-2.4.uci
Posted by roberts on Jan. 21 2008,16:26
Thanks to Curaga for:
gnucash.tar.gz
popt_dev.dsl


Thanks to Jason for:
xcdroast.unc

Thanks to Martin for:
pfe0.9.14.tar.gz

Read the info files for more information.

Posted by lucky13 on Jan. 21 2008,17:01
I finally remembered to submit ruby-1.8.6.uci. This has no symlink to irb and is a full, bloated package. I'll work on making it smaller in the very near future.
Posted by roberts on Jan. 22 2008,01:06
Quote (lucky13 @ Jan. 21 2008,09:01)
I finally remembered to submit ruby-1.8.6.uci. This has no symlink to irb and is a full, bloated package. I'll work on making it smaller in the very near future.

Thanks to Lucky13 for
ruby-1.8.6.uci

Posted by roberts on Jan. 23 2008,05:46
Thanks to Jason for:
xchat2.unc

Posted by stupid_idiot on Jan. 25 2008,02:54
Would anyone be interested in a chroot extension for building stuff?
This is the first time I am trying LFS. I don't plan to build 'everything', just a minimal set of packages for compiling.
-- linux-2.4.31 headers
-- glibc 2.3.2
-- same versions of libraries as in DSL
-- binutils 2.18
-- GCC 3.3.6
-- misc tools (autotools, bison, flex, libtool, m4, make, pkg-config, yacc, etc...)
-- perl 5.8.0
-- python 2.5.1
-- tcltk 8.4

Cheers!

p.s.
I had to do this sooner or later because I felt frustrated over some things with the Debian Sarge chroot.
Now I am compiling LFS on a Debian Woody chroot.
I set up a user called 'lfs', and I am following all the instructions.

If anyone has any suggestions I would love to hear them.
Thanks!

Posted by roberts on Jan. 25 2008,04:14
Thanks to Lucky13 for
gimp-1.2.5.uci

Posted by lucky13 on Jan. 25 2008,06:46
Quote
Thanks to Lucky13 for
gimp-1.2.5.uci

Can you please fix the md5sum link in the mirror? It's presently ..md5.txt instead of ..md5sum.txt. I don't know if that would cause md5sum failure via the MyDSL browser, but the correct txt page loads when I add "sum" to it (thought I saw it autocomplete).
< http://distro.ibiblio.org/pub....sum.txt >

Posted by curaga on Jan. 25 2008,22:30
S_I: Many people don't chroot, but if it makes things cleaner, go for it :)

Are you following a recent LFS with old packages (ie, some instructions and patches don't apply) or an old LFS?

Oh. Have fun, LFS is a great way to get a system just to your liking!

Posted by stupid_idiot on Jan. 25 2008,23:40
I am using LFS 5.0.
< Link to LFS-5.0 index > [archive.linuxfromscratch.org]
I'm sorry I forgot to mention which version I was using.

Re: patches
I only patched gcc and glibc so far.
For gcc, the patches are meant for gcc-3.3.1, but would still be relevant for gcc-3.3.6, so I manually edited a few files inside 'gcc-3.3.6/'. The patches are very small and very simple, so it's not very hard.

For glibc:
< glibc-2.3.2-sscanf-1.patch >
< glibc-2.3.2-inlining-fixes-2.patch >

For gcc pass 2:
< gcc-3.3.1-specs-2.patch >
< gcc-3.3.1-no-fixincludes-2.patch >
For gcc last pass (actual LFS install):
('gcc-3.3.1-specs-2.patch' from pass 1 must NOT  be applied this time.)
< gcc-3.3.1-no-fixincludes-2.patch >
< gcc-3.3.1-suppress-libiberty.patch >

For other software, e.g. coreutils, zlib, etc, I am using newer releases, so there is no need to apply the LFS-5.0 patches.

Posted by stupid_idiot on Jan. 31 2008,15:22
Small update:
I am no longer using '-fdata-sections -ffunction-sections' in $CFLAGS/CXXFLAGS and '-Wl,--gc-sections' in $LDFLAGS because '--gc-sections' does not work for all packages.
Am also not using '-fomit-frame-pointer', purely for the sake of convenience.
Now using simply CFLAGS/CXXFLAGS='-Os' for all packages in the experimental chroot-environment extension.

I am thinking of naming the extension 'dev-chroot.unc'.
Is that a good name??

Thanks for reading.

Posted by lucky13 on Jan. 31 2008,16:51
Note for slock.dsl:

I thought it locked out ctrl-alt-backspace, but that's apparently not the case. I think slock is better than nothing, but knowledgeable and/or crafty people can still get past it by killing X. I also think it's better than xscreensaver since it's a lot smaller in size and has lower resource demands (no daemon running in background) as long as the reason for running xscreensaver is to lock and/or require password for resuming a session in X. If you want a daemon to automatically set off fireworks or whatever after certain periods of inactivity, then slock never would appeal to you anyway.

If tighter security than this is desired, I can point users to gnu screen. If you're willing to suspend your X session (go to console) while in screen, you can use screen's lock (C-a C-x) to effectively keep users who don't have or can't break your password from using your computer (without rebooting). And if you're using console apps, you can resume your X session without ever interrupting any of your console apps running in screen. And, of course, screen allows you to detach and reattach sessions even remotely if you need to access your session(s) from elsewhere.

Posted by chaostic on Feb. 02 2008,19:03
Quote (stupid_idiot @ Jan. 25 2008,18:40)
I am using LFS 5.0.
< Link to LFS-5.0 index > [archive.linuxfromscratch.org]
I'm sorry I forgot to mention which version I was using.

Re: patches
I only patched gcc and glibc so far.
For gcc, the patches are meant for gcc-3.3.1, but would still be relevant for gcc-3.3.6, so I manually edited a few files inside 'gcc-3.3.6/'. The patches are very small and very simple, so it's not very hard.

For glibc:
< glibc-2.3.2-sscanf-1.patch >
< glibc-2.3.2-inlining-fixes-2.patch >

For gcc pass 2:
< gcc-3.3.1-specs-2.patch >
< gcc-3.3.1-no-fixincludes-2.patch >
For gcc last pass (actual LFS install):
('gcc-3.3.1-specs-2.patch' from pass 1 must NOT  be applied this time.)
< gcc-3.3.1-no-fixincludes-2.patch >
< gcc-3.3.1-suppress-libiberty.patch >

For other software, e.g. coreutils, zlib, etc, I am using newer releases, so there is no need to apply the LFS-5.0 patches.

If it helps in compiling for DSL, awesome.

I've been trying to find a a distro that had the same kernel as dsl, apt-get, gcc and gnu-utils while allowing usb boot and apt-getting to ram (Live cd compile only :D) but no luck :/

Even then, usb boot is only needed if I couldn't get live-cd toram and apt-get standard. Debian Sarge with a 2.4.27 kernel installed to a HD won't usb-boot :/

Posted by WDef on Feb. 03 2008,00:55
Debian Sarge has a newer glibc than dsl, unfortunately.  Woody is much closer.
Posted by stupid_idiot on Feb. 04 2008,10:29
IIRC, the version of < libc6 > in Sarge is the same(?) as DSL; but it was built on a machine with glibc-2.3.4 -- this can be verified by running './libc.so.6' on libc6 in the Sarge DEB.
I am guessing they compiled recent(?) versions of glibc on Etch (Etch uses glibc-2.3.4).
This seems to cause glibc-2.3.4-related symbols to appear in 'libc-2.3.2.so'.
This causes certain packages, when compiled, to give this error:
Code Sample
/lib/libc.so.6: version `GLIBC_2.3.4' not found (required by [binary])

It seems that, as a rule:
The version of glibc you use to compile a certain version of glibc should not be newer than the version of glibc on your target (e.g. DSL, other-distro).

Posted by yrogerg64 on Mar. 06 2008,21:34
Quote (curaga @ Jan. 20 2008,03:02)
Two extensions coming:

< Gnucash.tar.gz - gtk1 >
< Popt_dev.dsl >

All three files included in those (ext, md5, info)
(Gnucash was too big for my email, and popt-dev just came along)

edit: gnucash pack is not packed with bz2, seems I mistakedly used -z instead of -j with tar.

Thanks for the Gnucash app.  This one based on Gtk1 works seemlessly (and very fast I might add) on my tired old eTower 420i2 (Celeron 400 Mhz, 256 MB ram).  I had been running the latest Window$ build on the fat32 partition, but it would take approx. 2 minutes to load.  As an accountant, I very much appreciate having this extension working with DSL.

Posted by curaga on Mar. 07 2008,08:43
You're welcome :D
Posted by WDef on Mar. 08 2008,02:17
Quote
IIRC, the version of libc6 in Sarge is the same(?) as DSL


Looks like you were right stupid_idiot (I only just got back to this thread).  I never noticed these were both 2.3.2, only that Sarge binaries threw a glibc version error.

Sarge has glibc 2.3.2.ds1-22 < http://packages.debian.org/source/sarge/glibc >

DSL has libc6 2.3.2.ds1-10 (according to dpkg)

(but website package list says  2.3.1-16, out of date I suppose, probably where I got the idea dsl's glibc was older)

Woody has 2.2.5-11.8

But why would Debian build Sarge's glibc against a newer glibc6? Seems  nuts.

Posted by stupid_idiot on Mar. 10 2008,12:06
Hi WDef:
Seems like I was wrong. They did not build against a newer version of glibc. They only added a GLIBC_2.3.4 versioned symbol to the library. (I don't know why they do this, but I'm sure there must be a reason..)

From < Debian changelog >:
Quote
glibc (2.3.2.ds1-21) unstable; urgency=high
- debian/rules: Bump up shlib_dep_ver 2.3.2.ds1-21. It's required by adding GLIBC_2.3.4 symbol.
- debian/patches/sched-update.dpatch: Update sched_[gs]et_affinity to new interface and library version. Add GLIBC_2.3.4 versioned symbol for new interface. (Closes: #297769)
Bug #297769: < glibc: sched_setaffinity() provides obsolete interface >

Posted by lucky13 on Mar. 10 2008,15:00
Quote
(I don't know why they do this, but I'm sure there must be a reason..)

The bug tracking system at Debian explains why: "The current Sarge glibc still provides the obsolete 2-argument interface for sched_setaffinity(). As more software is starting to use this system call, this is becoming a real issue because developers will have to create a special version just for Debian."

Powered by Ikonboard 3.1.2a
Ikonboard © 2001 Jarvis Entertainment Group, Inc.