IRC chat logs for #ltsp on irc.freenode.net (webchat)


Channel log from 8 March 2008   (all times are UTC)

00:07primeministerp has quit IRC
00:07primeministerp has joined #ltsp
00:18tux_440volt has joined #ltsp
00:20tux_440volt has quit IRC
00:56
<warren>
cyberorg, hmm, I didn't think the fedora spec would be anywhere near usable on suse
00:56
<cyberorg>
warren, it worked after those changes :)
00:56
<warren>
cyberorg, I named some stuff "k12" thinking it would be Fedora/RHEL stuff, but if it works on SuSE then we need a different name.
00:57
"notdebian"? =)
00:57
<cyberorg>
i just worked on getting rpm building, i dont know what k12 things are
00:58
<warren>
cyberorg, another problem is that the more conditionals I have in the spec, especially non-fedora conditionals, the more likely it will be rejected by fedora package review.
00:58
cyberorg, some of your changes make a lot of sense, like s/i386/thatmacro/
01:00
<cyberorg>
warren, upstream rpm spec would have many distro conditionals, check out the spec files used on opensuse build service, they build packages for rhel, fedora, centos, mandriva, and debs for ubuntu and debian
01:01
<warren>
cyberorg, I was thinking we either need to rip out the spec from upstream (like debian ripped out the debian directory) or have multiple distro-specific specs
01:01
cyberorg, but if it REALLY does work with those simple changes then we can try to keep it unified.
01:02
<cyberorg>
at least they built the packages :)
01:02
<warren>
I'll first add some of the changes that make lots of sense
01:02
<cyberorg>
http://cyberorg.kicks-ass.org/ltsp/repodata/
01:04
<warren>
#
01:04
+ %if 0%{?suse_version}
01:04
#
01:04
+ install -m 0644 /usr/share/syslinux/pxelinux.0 $RPM_BUILD_ROOT%{_tftpdir}/ltsp/i386
01:04
this could be written with a conditional that looks for filename instead of suse macro
01:04
if [ -e /path/to/file]; then
01:05
<cyberorg>
yup, brb
01:05
<warren>
SYSLINUXDIR=
01:07
cyberorg, does suse's rpmbuild have a predefined %{dist} macro?
01:10
<cyberorg>
warren, what does it do?
01:11
<warren>
cyberorg, Fedora 8 packages have %{dist} defined to .fc8
01:11
.fc9
01:11
etc.
01:11
RHEL5 is .el5
01:11
<cyberorg>
wouldn't that be suse_version?
01:12
no we dont have dist
01:12
<warren>
hmm
01:12
maybe I'll get the package review done this weekend BEFORE the suse macros go in
01:12
<cyberorg>
having a dist makes rhel engineers lazy, they still have many fc rpms on rhel media
01:12
<warren>
so the reviewers don't notice
01:13
cyberorg, that's intentional
01:13
RHEL forks off from Fedora occasionally
01:13
and there is no reason to rebuild many of the packages
01:13
<cyberorg>
yeah, but at least they should change the dist :P
01:13
<warren>
Nah
01:13
no reason to do so
01:14
<cyberorg>
k :)
01:14
<warren>
#
01:14
+for arch in %ix86 x86_64 ppc ppc64; do
01:14
#
01:14
+ mkdir -p $RPM_BUILD_ROOT%{_tftpdir}/ltsp/$arch
01:14
Is this really a good thing?
01:15
you have tftp dirs named i586 and i686?
01:16
<cyberorg>
it gave error here as it was looking for i486 on my machine
01:17johnny has quit IRC
01:18
<warren>
cyberorg, you need to edit the scripts to redirect i486 i586 or i686 to i386
01:18
there's no good reason to have all archs have their own tftp directory
01:19
<cyberorg>
warren, yup we should just have i386, x86_64
01:19
<warren>
cyberorg, why does suse's RPM buildrequire both tftp and tftp-server?
01:20
cyberorg, does your tftp-server have /etc/xinetd.d/tftp too?
01:20
<cyberorg>
it doesnt, just tftp
01:20
yes, it is there
01:20deavid has joined #ltsp
01:20
<warren>
oh, I see.
01:20
<cyberorg>
we dont have tftp-server package
01:21
<warren>
you have a syslinux package?
01:21
<cyberorg>
yes
01:22
see how good multi distro rpm spec file look like :) https://build.opensuse.org/package/view_file?file=apache2.spec&package=apache2&project=Apache
01:23
<warren>
Fedora disallows complicated multi-distro spec files
01:23
I'll incorporate most of your changes
01:23
but we may have to fork after this weekend
01:23
especially because a lot of my scripts might not apply to you
01:23
the K12_DIST part totally doesn't apply to you
01:23
<cyberorg>
yeah
01:23johnny has joined #ltsp
01:25
<warren>
#
01:25
+for arch in %ix86 x86_64 ppc ppc64; do
01:25
#
01:25
+ mkdir -p $RPM_BUILD_ROOT%{_tftpdir}/ltsp/$arch
01:25
yeah, can't do this
01:25
because it would create too many directories
01:25
<cyberorg>
i had different spec file for each ltsp packages (ltspfs, ltsp-client, ldm) i'll see if i can make single one like this
01:25
<warren>
cyberorg, suse's base x86 RPM arch is i486?
01:25
no
01:25
<cyberorg>
may be you can i686
01:26
<warren>
cyberorg, I have different spec files for each of those because they each have their own source repository
01:26
cyberorg, if you build a 32bit x86 package what is the default arch?
01:26
i386?
01:26
<cyberorg>
i586 mostly
01:28
<warren>
cyberorg, our compiler guys did testing and we found that binaries built using i386 compatible instructions but i686 optimized are actually FASTER than i486 or i586 instructions on a i686 chip.
01:28
cyberorg, confusing huh?
01:29
cyberorg, very few of our packages are built not-i386. i586 and i686 kernel. i686 glibc, openssl. that's about it.
01:30
cyberorg, I would be focused helping me to fix things like ldm, because a change coming will break it on ALL distros.
01:30
cyberorg, (ldm uses X -ac which is totally wrong)
01:30
ldm never used x authentication
01:30
<cyberorg>
ftp://ftp-1.gwdg.de/pub/opensuse/repositories/server%3A/ltsp/openSUSE_10.3 packages are i586 and x86_64
01:31
ssh -X automatically sets up xauth?
01:31
<warren>
cyberorg, not in the way that we want
01:31
cyberorg, but something similar we will need to implement in ldm
01:32
cyberorg, cyberorg, after this weekend, I'm proposing that we create a ltsp-trunk/rpm directory that contains different spec files. Then mkdst will use your distro's spec file from that directory if $name.spec is missing from the base directory.
01:33
cyberorg, I just pushed a new ltsp.spec to ltsp-trunk. Go ahead and rebase your changes on top of that.
01:33
perhaps
01:33
ltsp-trunk/rpm/k12linux/ltsp.spec
01:33
ltsp-trunk/rpm/opensuse/ltsp.spec
01:36
cyberorg, along with that change I'm going to move more directories around and create new hierarchies
01:36
cyberorg, so it becomes more organized
01:36
cyberorg, so I would recommend forking your RPM spec after all the movement
01:36
<cyberorg>
that is the problem, lsb_release -i shows Distributor ID: SUSE LINUX (with space) vagrant did say something about doing tr " " "_"
01:37
<warren>
cyberorg, I recommend eliminating lsb_release
01:37
cyberorg, have your package hardcode a replacement string at RPM build time
01:38
lsb_release is one of upstream's bad ideas. I've completely got rid of that dependency from Fedora's LTSP
01:39
cyberorg, does suse use /etc/sysconfig/$foo files?
01:39
<cyberorg>
i guess we should remove all distro specific stuff out, and just have Makefile that installs stuff
01:39
yeah, we use that a lot, yast depends on it :)
01:41
<warren>
cyberorg, that is one possible option, but the Makefile would become the ugly one
01:41
cyberorg, and Makefile syntax is more difficult to work with than RPM syntax
01:42
I'm liking the idea of ltsp-trunk/rpm/$DISTRONAME/ltsp.spec
01:42
<cyberorg>
top level Makefile will be ugly, but it can be used to create rpms too
01:42
<warren>
cyberorg, have you been trying mkdst --testrpm?
01:42
<cyberorg>
yeah, that is what i used to create the packages you see on my server
01:42
<warren>
cyberorg, the entire point of mkdst is because ogra didn't want garbage like --testrpm in the Makefile
01:42
cyberorg, I can easily modify mkdst so it can optionally handle rpm/$DISTRONAME/ltsp.spec
01:43
<cyberorg>
make rpm, make deb could go in Makefile
01:43
<warren>
cyberorg, mkdst is supposed to support deb too one day
01:43
<cyberorg>
yeah mkdst --suserpm mkdst --fedorarpm
01:43
<warren>
cyberorg, another reason why we want to implement it in mkdst instead of the Makefile is you can have all the logic elsewhere and not duplicated
01:44
cyberorg, so you can use mkdst on 20 different source repositories
01:44
cyberorg, i'm not sure about --suserpm, might be a good idea
01:44
I'm also thinking about decoupling tagging from mkdst... because nobody uses it, and if people did use it the tool doesn't make it easy to use it correctly.
01:44
<cyberorg>
one mistake i made was copy ltsp-trunk/mkdst in bin/ and had my cup over 50% for a long time till i realized that i had to download another utility
01:45
*cpu
01:45
<warren>
You should make a mkdst RPM
01:45
you can use mkdst to make itself =)
01:45
Yeah, I think I'll start adapting mkdst to be cross-distro compatible tomorrow
01:45
at least both of us
01:45
cyberorg, please try the mkdst.spec and send me any changes you need to make it suse compatible.
01:46
<cyberorg>
ok, i'll get working on it right away
01:48
warren, everything looks ok, except make install DESTDIR=$RPM_BUILD_ROOT, we have %makeinstall macro for that
01:48
but that is fine too
01:49
<warren>
cyberorg, %makeinstall doesn't work consistently so I hard coded it in that spec just to be safe.
01:49
<cyberorg>
k, it is fine
01:54
warren, is mkdst only for ltsp of any VCS/bzr project?
02:02
cp -pr README-release.conf COPYING Changelog /var/tmp/mkdst-0.0.0_080308-1-root-root/usr/share/doc/packages/mkdst
02:02
cp: cannot stat `Changelog': No such file or directory
02:02
warren, you need to touch Changelog
02:10daduke_ has joined #ltsp
02:14Q-FUNK has joined #ltsp
02:25captain_magnus has joined #ltsp
02:37captain_magnus has quit IRC
02:37captain_magnus has joined #ltsp
02:38
<cyberorg>
captain_magnus, are you going to stick around?
02:38
<captain_magnus>
cyberorg: Today? For a bit...
02:39
<cyberorg>
i meant you dropped of twice in a minute from the channel :)
02:39
<captain_magnus>
Yeah, made a few changes to xchat which required a restart :-)
02:48
<cyberorg>
warren, here are the spec files i am using, let me know what changes are required to build on fedora http://cyberorg.kicks-ass.org/ltsp/spec/
03:44Q-FUNK has quit IRC
04:33sepski has joined #ltsp
04:56daduke_ has quit IRC
05:05tux_440volt has joined #ltsp
05:05sepski has quit IRC
05:08sepski has joined #ltsp
05:30ogra_cmpc has quit IRC
05:47ogra_cmpc has joined #ltsp
06:08makghosh has joined #ltsp
06:16markit has joined #ltsp
06:17
<markit>
hi, forgive my ignorance, but I have a virtual machine that exports screen through VNC, and would like to have a thin client to connect with it
06:17
is it possible setup ltsp in a way that the thin client runs vnc instead of kdm/gdm?
06:17
I mean, that connects to a certain vnc screen instead of regular X-Window?
06:26plamengr has joined #ltsp
06:28tux_440volt has quit IRC
06:30Q-FUNK has joined #ltsp
06:48elisboa has joined #ltsp
07:23elisboa has quit IRC
07:29daduke_ has joined #ltsp
07:34J45p3r has joined #ltsp
07:40
<mnemoc>
markit: i don'T know much about ltsp yet, but take a look into http://www.karlrunge.com/x11vnc/
07:59
<stgraber>
markit: you certainly can write a LTSP screen script to connect to your vnc server
07:59
markit: you'll need to install vncviewer in the chroot, write the script and set a screen to use it (using lts.conf)
08:09makghosh has quit IRC
08:15makghosh has joined #ltsp
08:32allerbest has joined #ltsp
08:32
<allerbest>
Hi!
08:32
I've got a problem with LTSP & Edubuntu 7.10.
08:33
I created an environment for i386 on a amd64 machine.
08:33
Clients boot the kernel well, until ...
08:33
http://www.it42.de/temp/edubuntu_pxe_1.png
08:33
this state.
08:33
But the root path mentioned there is filled with bin, sbin, etc, var, usr etc.
08:34
Nothing suspicious in syslog. Any hints where to look for problems?
08:34
(sorry, I'm not a native speaker)
08:34mcfloppy has joined #ltsp
08:35
<laga>
that's an interesting error
08:35
allerbest: so /root exists and /root/sbin/init exists as well?
08:36
<allerbest>
you mean the /root as a client? This maps to /opt/ltsp/i386.
08:36
root@lhserv:~# ls -la /opt/ltsp/i386/sbin/init
08:36
-rwxr-xr-x 1 root root 88672 2007-09-17 04:14 /opt/ltsp/i386/sbin/init
08:37
(superfluous -a switch, please ignore :-)
08:37
<laga>
no, on the client itself. when it shows you the initramfs prompt
08:37
<allerbest>
ah, mompl
08:38
/root is there, containing 'cow' and 'rofs'
08:38
<laga>
allerbest: on the server, do you have /opt/ltsp/images/i386.img ?
08:39
<allerbest>
/sbin is there, containing swapon, usplash_write, pkill and so on, but not 'init'
08:39
<laga>
and there's no /root/sbin/ ?
08:40
<allerbest>
laga: root@lhserv:~# ls -l /opt/ltsp/images/i386.img
08:40
-rwxr--r-- 1 root root 146919424 2008-03-06 15:28 /opt/ltsp/images/i386.img
08:40
<laga>
allerbest: do you have a corresping entry for nbdrootd in /etc/inetd.conf ?
08:41
<allerbest>
See <http://www.it42.de/temp/edubuntu_pxe_2.png>
08:41
mompl
08:42
inetd.conf:
08:42
2001 stream tcp nowait nobody /usr/sbin/tcpd /usr/sbin/nbdrootd /opt/ltsp/images/i386.img
08:42
Seems to be fine.
08:42
<laga>
what's running in port 2001?
08:42
err
08:42shogunx has quit IRC
08:42
<laga>
2000
08:42
sorry
08:43
<allerbest>
laga: 2000 stream tcp nowait nobody /usr/sbin/tcpd /usr/sbin/nbdrootd /opt/ltsp/images/amd64.img
08:43
laga: But I renamed the amd64 path in /opt/ltsp because I only have i386 clients
08:43
I didn't know where this is anchored, so I choose this bad way to disable amd64
08:44
Is it better to comment the port 2000 line out?
08:44
<laga>
i'm a bit stumped and i don't remember the exact details how LTSP worked out the correct port & pxe boot path..
08:44
allerbest: i'd remove the 2000 line completely and let the i386 nbdrootd run on port 2000
08:44
maybe that helps
08:46
<allerbest>
I'll try that, mompl
08:47
how do I reload the (x)inetd after modifying the inetd.conf? Or is it read on the fly?
08:47shogunx has joined #ltsp
08:47
<laga>
sudo /etc/init.d/openbsd-inetd restart
08:47
something like that
08:48
<allerbest>
thanks
08:48* allerbest was looking for /etc/init.d/inetd or xinetd, but not for openbsd-* :-)
08:48
<allerbest>
Ha!
08:48
<laga>
been there, done that :)
08:48
<allerbest>
It is booting!
08:48
<laga>
yay
08:49
<allerbest>
Jabbadabbaduuu!
08:49
laga: Thanks a lot, YMMD!
08:49
<laga>
yeah, it was probably talking to port 2000 and didn't get a nice squashfs there
08:49
np :)
08:57daduke_ has quit IRC
08:58
<allerbest>
laga: Now installing all required applications. Thanks again & bbl!
08:58allerbest has quit IRC
09:25rage7 has joined #ltsp
09:25
<rage7>
hi
09:32rage7 has quit IRC
09:33tux_440volt has joined #ltsp
09:43makghosh has quit IRC
09:46makghosh has joined #ltsp
09:52topslakr has quit IRC
09:55Faithful has quit IRC
10:06Q-FUNK has quit IRC
10:13spectra has joined #ltsp
10:38topslakr has joined #ltsp
10:57daduke_ has joined #ltsp
11:01vagrantc has joined #ltsp
11:04
<vagrantc>
ok, so i've got all sorts of patches to upload :)
11:21daduke_ has quit IRC
11:21ogra_cmpc has quit IRC
11:38ogra_cmpc has joined #ltsp
11:41soneyka has quit IRC
11:41soneyka has joined #ltsp
11:48tux_440volt has quit IRC
12:35
<vagrantc>
otavio: who was it i should talk to regarding mips builds for ltsp and ldm ?
12:36
<otavio>
vagrantc: tbm, aurel32 or ths at #debian-boot on oftc
12:40ogra_cmpc_ has joined #ltsp
12:42* ogra_cmpc curses apt-proxy
12:42
<laga>
apt-cacher?
12:43
<ogra_cmpc>
nope, apt-proxy
12:43
<laga>
yeah, i was suggesting apt-cacher :) i have used it a few times lately and it seems to work fine here..
12:43
<ogra_cmpc>
does it re-use an existing cache ?
12:43
(an apt-proxy one)
12:44
<laga>
you can probably configure it to do that.. let's see
12:44
<ogra_cmpc>
i'm not really after rebuilding 10G of packagecache
12:45
<laga>
Q: Can I just copy some .debs into the cache dir and have it work (precaching)?
12:45
A: Almost! A bit additional work is also required to make them useable and persistent in the cache.
12:45
<ogra_cmpc>
hmm, i nfs mounted the cache for the other server locally ... running apt-proxy on localhost seems to work now
12:45
<laga>
looks like it'd work
12:45
ah, good
12:45
<ogra_cmpc>
s/for/from/
12:45* ogra_cmpc wasted nearly the whole day on that crap ... sigh
12:46
<laga>
damn, you could have uploaded LTSP in that time ;)
12:47
<ogra_cmpc>
i would be so happy if i had time for even looking at one line ltsp code
12:47
<laga>
heh
12:47* ogra_cmpc needs to have some of the classmate stuff ready by monday
12:47* laga is currently finishing the GUI for mythbuntu-diskless
12:47
<laga>
ogra_cmpc: good luck
12:47
<ogra_cmpc>
same to you
12:48
even though i suppose you dont have a deadline ... so yours might rather be fun
12:48
<laga>
true
12:50
<ogra_cmpc>
ah, at least the openoffice packages are a third smaller in hardy ...
12:50* ogra_cmpc hugs dpkg with lzma
12:50
<laga>
ooh, nice
12:51* laga wants squashfs with lzma
12:51
<ogra_cmpc>
yeah, since two weeks or so
12:52
lzma has a fat CPU overhead while decompressing ... not sure you would be happy with lzma squashfs
12:52
its trading speed for compression rate
12:52
<laga>
true, but all hose live cd guys seem to be happy with it
12:52
s/hose/those/
12:52Egyptian[Home] has joined #ltsp
12:53
<ogra_cmpc>
where ?
12:53
<laga>
my clients will usually be a bit fatter than the typical LTSP client
12:53* ogra_cmpc wasnt aware there was a kernel module with lzma support
12:53
<laga>
oh
12:53
<ogra_cmpc>
i know that wont be in ubuntu
12:53
<laga>
no, not in ubuntu
12:53
http://www.squashfs-lzma.org/
12:53
<ogra_cmpc>
ah
12:54
the prob is that the upstream kernel guys want to only support a single compression system
12:54
<laga>
yeah, who needs more than 650kb anyways.
12:54
s/650/640/
12:56
<ogra_cmpc>
you can make billions with such sentences :P
12:56
<laga>
yay, i'm gonna be rich
12:56
<vagrantc>
otavio: although, i have to make a new LDM upload fixing an RC bug anyways ...
12:56Egyptian[Home] has quit IRC
12:56
<ogra_cmpc>
vagrantc, did you happen to get around testing ldm without the -ac ?
12:57
<vagrantc>
ogra_cmpc: yes, i just posted to ltsp-developer about it.
12:57
<ogra_cmpc>
got no mail access here
12:57
does ltspfs work ?
12:57
thats the only thing i could imagine to break
12:57
<vagrantc>
ogra_cmpc: basically, drop -ac and drop xauth. ltspfs works, sound works.
12:57
<ogra_cmpc>
drop xauth ???
12:57
<vagrantc>
ogra_cmpc: i also included in the patch to add -ac with LDM_DIRECTX...
12:58
ogra_cmpc: it's not using xauth anyways.
12:58
<ogra_cmpc>
huh ?
12:58
<vagrantc>
ogra_cmpc: it's *never* used xauth, because we've been using -ac, which supercedes xauth.
12:58
<ogra_cmpc>
i get the cookie stored in roots Xauthority with xauth list
12:58
<vagrantc>
you sure do.
12:58
<ogra_cmpc>
so its using xauth
12:58
it never took effect because of -ac
12:58
<vagrantc>
no, you just have an un-used cookie in root's Xauthority file
12:59
<ogra_cmpc>
but it will if we drop -ac
12:59
<vagrantc>
and if you try to use it, it breaks.
12:59
<ogra_cmpc>
hmm
12:59
<vagrantc>
having TESTED this, i tell you drop xauth.
12:59
<ogra_cmpc>
oki
12:59
i will
12:59
<vagrantc>
so here's the part i'm unsure about ...
13:00
it gives a message somewhere along the line of "no valid xauth data found, generating fake data" ... what i don't know is if that xauth data is predictible and thus insecure... or randomly generated and not stored anywhere and thus, very secure.
13:01
but it does prevent people from writing/reading to the display
13:01
<ogra_cmpc>
do you get it as well if you keep xauth in ?
13:01
<vagrantc>
so it's certainly a step up, if still flawed.
13:01
<ogra_cmpc>
(i dont)
13:01
<vagrantc>
it breaks if i try to keep xauth
13:01
<ogra_cmpc>
could that be readonly /root ?
13:01
<vagrantc>
no, i've been testing with /root rw
13:01
<ogra_cmpc>
i can write to that
13:02makghosh has quit IRC
13:02
<vagrantc>
long-term, i thikn we just have to figure out xauth properly ... because we'll need that for local apps and such
13:02
but for now, i have a few line patch which seems to at least improve, if not fully fix the issue.
13:03
<mnemoc>
warren said something about a conflict on how ldm and xauth works
13:04
<ogra_cmpc>
mnemoc, fedora changed the mechanism between teh releases apparently
13:04
works in the lqast version but doesnt in the devel
13:04
<ltsppbot>
Someone pasted "only enable "-ac" in directx mode, comment out call to create_xauth" (22 lines) at http://pastebot.ltsp.org/461
13:04
<vagrantc>
that was me
13:04* ogra_cmpc waves to someone
13:05
<vagrantc>
that patch works for me ...
13:05
<ogra_cmpc>
cross your fingers i'll not run out of ram ...
13:05
<vagrantc>
that's from current ldm-trunk ... but i think it should apply to the version in debian or ubuntu
13:05* ogra_cmpc starts a browser
13:05Egyptian[Home] has joined #ltsp
13:06
<mnemoc>
do you know if "upstream" plans to make a tarball release on day?
13:07
<ogra_cmpc>
mnemoc, xorgh ?
13:07
they do
13:07* vagrantc wonders if LTSP will ever make an upstream release
13:08
<ogra_cmpc>
well, thats what mkdst is for ... and we should set a version at least
13:08
so at least branching off something into a stable branch or so should hapopen
13:08
<mnemoc>
and tagging the versions :p
13:08
<ogra_cmpc>
vagrantc, loks good to me, even though i'd like to find out why xauth doesnt work
13:09Demon has joined #ltsp
13:09
<vagrantc>
well, apparently ltsp5 has been handed over to the distros ... so should we just say, for now, that fedora + debian + ubuntu is the ltsp cabal and we can collectively make a release happen? :)
13:09
<ogra_cmpc>
mnemoc, thats not necessary if you branch off
13:09
right
13:09
<vagrantc>
ogra_cmpc: i'd like to figure it out too, but i also really just want it fixed.
13:09
<ogra_cmpc>
as long as ltspfs and sound keep working in default mode i'm happy
13:09
<vagrantc>
that's just a huge security flaw that i'm ashamed i didn't fight harder about years ago.
13:09
<ogra_cmpc>
directx is unsafe as hell anyway
13:10
yeah
13:10Demon is now known as Vorlock
13:10
<ogra_cmpc>
what bothers me most is that it was carried over from python
13:10
<mnemoc>
ogra_cmpc: if one forks it's own branch you need to invest much more work than just trusting upstream, and most non-major distributions can't afford that
13:10
<ogra_cmpc>
so we all could have seen it
13:10
<Vorlock>
hi
13:11
<ogra_cmpc>
mnemoc, *we* (upstream) will fork off a brach we will call stable ... that branch wont change then
13:11
<mnemoc>
ogra_cmpc: great :)
13:11
<ogra_cmpc>
thats what i mean with forking :)
13:11
the -trunk branches were supposed to be moving targets
13:12
<mnemoc>
ogra_cmpc: that's exactly we i fear to use it :p
13:12
s/we/why/
13:12daduke_ has joined #ltsp
13:12
<ogra_cmpc>
if we all agree we think its stable we'll branch off a snapshot (ltsp-5.2 or so )... development will go on in trunk
13:12
mnemoc, who is we ?
13:13
<mnemoc>
ogra_cmpc: it was a typo, but i want to integrate ltsp5 in opensde.org, a framework to build and maintain custom distributions for multiple architectures
13:14
<ogra_cmpc>
ugh
13:14
thats a challenge
13:14
ltsp5 relies pretty much on stable policies and rules the distro sets ...
13:15
doing that on a moving target that can build custom stuff is very tricky i guess
13:17
<mnemoc>
ogra_cmpc: that's why prefer to "trust" upstream, and minimize patching, so we (a very small comunity) can focus on making the agnostic integration easy for poeple using opensde to maintain their specilized distributions
13:17
<Vorlock>
where you download ltsp5 ? on main page (ltsp.org) last version is 4.2
13:17
<mnemoc>
Vorlock: https://launchpad.net/~ltsp-upstream
13:17
<ogra_cmpc>
mnemoc, well, look at the code its exactly built for doing your own thing through distro specific plugins
13:18
<mnemoc>
ogra_cmpc: yes, v5 is very friendly in that sense
13:18
<ogra_cmpc>
Vorlock, ltsp5 is supposed to be implemented by your distro
13:18
<Vorlock>
mnemoc thnks
13:18
<ogra_cmpc>
Vorlock, what are you using ?
13:20
<Vorlock>
ogra_cmpcwhere i'm working we using asplinux11.2 and asplinux 12 (carbon) & ltsp 4.2 (tried)
13:20
but i have some questions on ltsp :(
13:21
<ogra_cmpc>
Vorlock, there is no ltsp5 version for rpm based distros yet, fedora is working ard on getting it there and there is a beta version somewhere (warren might be able to tell you where)
13:21
(seeimg asp is rpm based)
13:22
*seeing
13:22
<Vorlock>
i'm from Ukraine :) my English not very well as you :)
13:23
<mnemoc>
i'm more exotic, i'm from chile :)
13:23* ogra_cmpc is german
13:24
<mnemoc>
ogra_cmpc: most people in free software seems to be german
13:24
<Vorlock>
Asplinux 12 based on FC7, if ltsp5 work on FC then work on ASP12, may be...
13:25
<ogra_cmpc>
Vorlock, its supposed to be in the next fedora release
13:25
Vorlock, wait for warren he has packages
13:26
<Vorlock>
Germany - my blue idea :)
13:26
13:26
sorry, hight dream
13:27
<ogra_cmpc>
mnemoc, european would probably be true ... :) i think there are more OSS devs in the uk
13:30
oooh, yippie ... finally the first apt-proxy run that works i 6h of fiddling
13:30
<Vorlock>
next version it's good, but my corporation is working only asp :( & what are you using ?
13:30
<ogra_cmpc>
ubuntu
13:30
<mnemoc>
ogra_cmpc: really? at least in IRC i haven't met any brit yet
13:31
<ogra_cmpc>
i do that all the time :)
13:31
<mnemoc>
:)
13:31
<ogra_cmpc>
(many of my colleagues are brits)
13:33
<Vorlock>
in you company using free software? not MS ?
13:34
<ogra_cmpc>
we're writing free software :)
13:34* mnemoc assumes ogra_cmpc works for canonical
13:34
<stgraber>
ogra_cmpc: really ? I would have said that most UK based people from Canonical were more doing management stuff and tech guys would be mainly Germany/US based (at least for people I work with)
13:35
<ogra_cmpc>
stgraber, the distro team was 1/3 german 1/3 uk and 1/3 rest of the workld for quite some time
13:35
that changed with the recent growth of the company
13:35
there are more us based people now
13:35
<Vorlock>
US it's Unated States ?
13:35
<ogra_cmpc>
yes
13:35
<Vorlock>
thnks
13:41
<stgraber>
btw, as we already are off-topic :) Does one of you know how to change contrast/brightness of a laptop LCD ? (I already know xgamma and it doesn't help)
13:41
<mnemoc>
ogra_cmpc: when do you think an "stable" branch of ltsp{,fs} and ldm may appear? days? weeks? months?
13:42
<vagrantc>
there are many stable branches :)
13:42
<ogra_cmpc>
mnemoc, well, for ubuntu we have stable sources since 2.5 years
13:42
debian since about 1.5
13:42
<mnemoc>
ogra_cmpc: i mean from ltsp itself
13:42
<vagrantc>
mnemoc: the main thing is what do you mean by "stable"
13:42
<ogra_cmpc>
right
13:43joebaker has joined #ltsp
13:43
<ogra_cmpc>
stable means properly integrated with a distro
13:43
else it cant be satble
13:43
<vagrantc>
there has been no release of ltsp from ltsp.org since 4.2
13:43
<ogra_cmpc>
*stable
13:44
<mnemoc>
ogra_cmpc: in this case i mean "if the proper integration with the distro is implemented, it will work fine"
13:44
ogra_cmpc: -trunk has at least the xauth problem with is distro-independent
13:44
which*
13:44* Vorlock reading documentations
13:44
<vagrantc>
!integration
13:44
<ltspbot`>
vagrantc: "integration" is http://wiki.ltsp.org/twiki/bin/view/Ltsp/IntegratingLtsp
13:45
<vagrantc>
that's got some old URLs, especially for bzr branches ... but the basic idea is there.
13:47
<mnemoc>
vagrantc: i'm doing that (actually doing bzr integration first), i'm talking about an "stable" upstream i can use (without branching) as base for my integration
13:48
<vagrantc>
mnemoc: and i'm telling you that upstream hasn't made any mention of making an upstream release any time in the near future.
13:48
it's basically left to the distros
13:48
<ogra_cmpc>
you wont be able to use it without branching
13:48
<mnemoc>
:\
13:48
<ogra_cmpc>
branch off, add your plugins and give it to us to merge
13:48
<mnemoc>
that's not very helpful for distributions without tons of man-power
13:49
and ltsp claims to have the goal to be integrated on more than just major distributions
13:49
<vagrantc>
and a stable release will likely require patches extensive patches if the distro wasn't well supported at the time of release...
13:49
<ogra_cmpc>
well, the core concept of ltsp5 is that you install ltsp-server with your package manager and that ltsp-build-cdlient sets up a client environment for you using your distros package management software and configuration tools ...
13:50
<mnemoc>
vagrantc: the thing of the stable upstream is mostly "if don't work , it's my fault" and not "it don't work, is upstream broken?"
13:50
<vagrantc>
mnemoc: well, you can take the bzr branch from today and call it a release ... there's no way to support a release for your distro without integrating it into your distro.
13:50
<ogra_cmpc>
*i* dont know your package manager, neither do i know confg policies etc
13:50
so its up to a distro person to integrate it
13:51
<mnemoc>
you insist in mixing things up :\
13:51
<vagrantc>
there is no mixing. ltsp5 is based entirely on the distro it's being made for.
13:51
there isn't much to release.
13:51
that's the point ... ltsp5 is 99% $DISTO and 1% ltsp.
13:52
<daduke_>
vagrantc!!!
13:52
<mnemoc>
not enterely, ldm and ltspfs _are_ common across distributions, and deserve releases
13:52
<vagrantc>
fair enough.
13:52
mnemoc: why don't you ask upstream to make a release, then.
13:53
<ogra_cmpc>
mnemoc, they usually depend on features in the other youcant just divide them apart
13:53
<vagrantc>
ogra_cmpc: yes, but that's improving.
13:53
<ogra_cmpc>
ltspfs currently doesnt work without ldm for example
13:54
<vagrantc>
anyways, gotta run ... be back in a few to make a new upload of ldm and possibly ltsp ...
13:54vagrantc has quit IRC
14:02
<mnemoc>
ogra_cmpc: i guess after all these years (8) working on stuff to help to make distributions i deformed myself to value more what's common across all than what makes them different... i general i have troubles discussing with distribution-centric people :(
14:02
<ogra_cmpc>
mnemoc, the common parts in ltsp5 are only a few k
14:03
look at the code
14:05
the core parts are the builder script and the initscripts in the client environment ... both are totally depending on what your distro uses
14:06plamengr has quit IRC
14:07bobby_C has joined #ltsp
14:07
<mnemoc>
ogra_cmpc: what's better to use as reference -mainline or -trunk ?
14:07
<ogra_cmpc>
-trunk
14:07
where did you find mainline ?
14:08
<mnemoc>
http://wiki.ltsp.org/twiki/bin/view/Ltsp/IntegratingLtsp
14:08
<ogra_cmpc>
thats ancient (vagrant mentioned that)
14:09
nobody updated the wikipage in a while
14:09
<mnemoc>
i noticed :p
14:09
it's 1y old
14:10
<ogra_cmpc>
well, until about 6 months ago it was valid
14:10
then we started to split the code into sparate branches
14:11
and founded the ltsp-upstream team etc
14:11
<mnemoc>
ic
14:13
<ogra_cmpc>
(preyy much when fedora entered the development) ... mainline was valid for the two years before though
14:13
*pretty
14:15
<mnemoc>
is warren the only developer from fedora?
14:16
<ogra_cmpc>
well, he has the whole redhat crew in the back :)
14:16
so he can delegate tasks
14:16
<mnemoc>
:)
14:16
<ogra_cmpc>
but he's currently the only one doing constant work on the fedora implementation
14:22daduke_ has quit IRC
14:29
<Vorlock>
who use ltsp 4.2 or 4.0 ?
15:00Vorlock has quit IRC
15:38joebaker has quit IRC
15:39ogra_cmpc has quit IRC
16:30vagrantc has joined #ltsp
16:31ogra_cmpc has joined #ltsp
16:39ogra_cmpc has quit IRC
16:48
<warren>
This Alt Linux guy is rude
16:48gonzaloaf has quit IRC
16:50
<rjune>
warren: ?
16:50
<warren>
or maybe he's just a user
16:50
I don't know who he is
16:50
<rjune>
what alt linux guy?
16:51gonzaloaf has joined #ltsp
16:51
<warren>
michael shigorin
16:52
and somebody else who wrote something nonsensical on my blog
16:52
might be the same person
16:52
<rjune>
ah
17:00bobby_C has quit IRC
17:00
<vagrantc>
warren: yeah, i've always sort of thought of them as just shy of a troll.
17:00
sometimes actively trolling
17:01
i'm about to reply to a seemingly reasonable email, with the slight fear that i might get a bit of trolling for it.
17:01
<warren>
altlinux is another RPM based distro?
17:02
what are they supposed to be good at?
17:02
<vagrantc>
warren: regarding the patches for ldm ... i wasn't intending to apply the patch to ldm-trunk, just getting out there a *something* that fixed the security issue while we figure out the long-term issue with xauth
17:02
warren: it's apparently being deployed in ... russian schools ?
17:12markit has quit IRC
17:37
<vagrantc>
so, ldm has dropped support for selecting anything that's not a UTF-8 locale?
17:48ogra_cmpc has joined #ltsp
17:49ogra_cmpc has quit IRC
17:50gonzaloaf has quit IRC
17:53gonzaloaf has joined #ltsp
18:02
<vagrantc>
warren: can we conditionalize the Xsession handling at build time?
18:03
with the current code, it takes two filesystem calls on debian with what previously only did one...
18:17sepski has quit IRC
18:19milesd has quit IRC
18:19milesd_ has joined #ltsp
18:20milesd has joined #ltsp
18:20
<milesd>
Hi, everybody
18:20
I'd love to get involved with this project and start hacking but I seem to be at a slight loss understanding how it's organized
18:21
I found the bzr repositories on launchpad... actually somebody in here told me about them
18:21
but the ltsp-trunk repository doesn't appear to have code in it... or at least nothing I can build
18:21
is there some kind of "getting started guide" somewhere? Can any of you give me some pointers?
18:36
<mnemoc>
milesd: these are the four "upstream" branches https://code.launchpad.net/~ltsp-upstream
18:37
milesd: and this is the other interesting link i got when i asked http://wiki.ltsp.org/twiki/bin/view/Ltsp/IntegratingLtsp
18:37
<milesd>
mnemoc, yeah, I found that
18:37
mnemoc, aaah, the second link looks like some good reading, thank you :)
18:38
<mnemoc>
milesd: bzr get <url> ... even if you don't see code on the browser, bzr will find it
18:38
<milesd>
mnemoc, ah, I was using bzr branch...
18:39
mnemoc, I'm pretty familiar with mercurial, I'm starting to understand that bzr isn't very similar
18:39
<mnemoc>
i think get and branch are aliases of the same command
18:39
<milesd>
ah, ok
18:39
<mnemoc>
bzr clone seems to work too
18:39
https://code.launchpad.net/~ltsp-upstream/ltsp/ltsp-trunk <---- look at the "example" line
18:40
<milesd>
mnemoc, yeah, I've got the code... such as it is, I just don't understand what I'm looking at
18:40
mnemoc, why don't I read this "integration" page and get back to you?
18:41
<mnemoc>
milesd: sure, but i don't know much about ltsp neither
18:41
milesd: but take into account that integration page is outdated, but it's good reading anyway
18:42
<milesd>
mnemoc, yeah, at least it gives me a baseline
18:53toscalix has joined #ltsp
19:11toscalix has quit IRC
19:13
<vagrantc>
milesd: what distro are you working with?
19:20ogra_cmpc has joined #ltsp
19:29ogra_cmpc has quit IRC
19:34
<warren>
vagrantc, please don't make me optimize before it is fully implemented...
19:35
vagrantc, yeah, I agree let's do that, and SuSE even has a 3rd location for it.
19:39
<vagrantc>
warren: yeah, exactly.
19:40
warren: i don't quite understand the optimize comment though...
19:40
actually, might be a good thing to provide in ldminfod
19:41
but that's a little more code tweaking...
19:41
warren: you're talking about the default Xsession script location, right?
19:46milesd has quit IRC
19:47ogra_cmpc has joined #ltsp
19:53
<vagrantc>
gah. i updated virtualbox and it seems like X isn't configuring properly ... maybe.
19:53
just when i try to test a new upload, too.
20:01
oh... the new ldm i'm working on is also dependent on a new ltsp ... with all the /usr/lib -> /usr/share stuff
20:04
<ogra_cmpc>
indeed
20:10
<warren>
let's call the current ldm (after xauth fix) ldm-2.0 branch
20:10
new ldm with updated ldminfod protocol will be ldm-3.0
20:10
well....
20:10
ldm-trunk
20:15Egyptian[Home] has quit IRC
20:16Egyptian[Home] has joined #ltsp
20:19
<vagrantc>
warren: is the updated ldminfod protocol stuff actually implemented yet, or just an idea at this point?
20:21
hmmm... i thik i can patch the current ldm upload to work with both locations, and then when the ltsp + ldm hit lenny/testing, i can upload a new ldm without those changes
20:21
and new ltsp
20:22* vagrantc strains to think of a simple name for this patch
21:01ogra_cmpc has quit IRC
21:05
<warren>
vagrantc, a short list of things we need
21:15
<mnemoc>
http://wtogami.livejournal.com/23648.html <-- nice post
21:16
(it's about LTSP5 and fedora, not spam)
21:17
warren: is it yours?
21:22ogra_cmpc has joined #ltsp
21:33
<vagrantc>
well, new ldm upload to debian. hopefully this will actually get through.
21:34* vagrantc waves
21:34vagrantc has quit IRC
21:41ogra_cmpc_ has joined #ltsp
21:41ogra_cmpc has quit IRC
21:41ogra_cmpc_ is now known as ogra_cmpc
21:48ogra_cmpc has quit IRC
21:51J45p3r has quit IRC
21:55ogra_cmpc has joined #ltsp
21:57robbie has joined #ltsp
22:00
<dberkholz>
warren: i agree with you that actual released tarballs would be nice
22:01* mnemoc happy to read that
22:12ogra_cmpc has quit IRC
22:12ogra_ has quit IRC
22:13ogra_ has joined #ltsp
22:26ogra_cmpc has joined #ltsp