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


Channel log from 25 February 2009   (all times are UTC)

00:04try2free has joined #ltsp
00:08irule has quit IRC
00:14try2free has quit IRC
00:29pasmen has joined #ltsp
00:30pasmen has joined #ltsp
00:31Egyptian[Home] has quit IRC
00:31nothingman has joined #ltsp
00:32Egyptian[Home] has joined #ltsp
00:35w0lter has joined #ltsp
00:36w0lter has joined #ltsp
00:36w0lter has left #ltsp
00:55Daemonik has quit IRC
01:02nothingman has quit IRC
01:05hanthana_ is now known as hanthana
01:38phantom has joined #ltsp
01:38F-GT has quit IRC
02:00try2free has joined #ltsp
02:02alkisg has joined #ltsp
02:08
<Appiah>
O_o
02:14alkisg has quit IRC
02:49polytan has joined #ltsp
02:54aledec has quit IRC
02:54aledec has joined #ltsp
02:55alkisg has joined #ltsp
02:56wigwam has joined #ltsp
03:01_wigwam_ has quit IRC
03:01Gadi has quit IRC
03:02
<alkisg>
I got a client that worked fine until yesterday, and today it stops at "Starting hardware abstraction daemon hald". Any clues?
03:02Gadi has joined #ltsp
03:04evilx_ has quit IRC
03:07Nubae has joined #ltsp
03:07ogra has quit IRC
03:08ogra has joined #ltsp
03:18try2free has quit IRC
03:35mikkel has joined #ltsp
03:50Nubae has quit IRC
03:52hanthana_ has joined #ltsp
03:57
<polytan>
hi
03:57
do you know how to activate 3D acceleration on my thin clients to be able to watch a video without seeing frames ?
03:59artista_frustrad has quit IRC
04:02alkisg1 has joined #ltsp
04:03hanthana has quit IRC
04:14artista_frustrad has joined #ltsp
04:15
<alkisg>
polytan: it depends on the graphics cards, what do your clients have?
04:15
(and video would be xv, not 3d)
04:16
<polytan>
hum
04:16
I've got nvidia cards, ati cards, intel ones and sis
04:16
<alkisg>
...and you lose frames in all of them?
04:17
E.g. for nvidia, you'd have to install the nvidia drivers into the chroot. Intel should work out of the box...
04:18
...and of course you'd need a fast network, e.g. maybe 50mbps per client watching video...
04:19
<knipwim>
i lost a lot of frames in my X over ssh
04:20
which was definitly lack of cpu
04:21alkisg1 has quit IRC
04:21
<Appiah>
polytan install the correct driver and force som X settings
04:22
<polytan>
ok
04:22
I saw a huge differente between 100Mbits and Gibagit network
04:22
it is amazing
04:22
especially on 14 machines
04:26alkisg has quit IRC
04:31
<Appiah>
polytan: what kind of things does the clients do?
04:33
<polytan>
lot of things
04:33
but when watching a video, it is quite slow
04:33synchris_odium has joined #ltsp
04:33
<polytan>
I can apreciate each picture when watching a movie ;)
04:34
<knipwim>
do you use x over ssh?
04:35alkisg has joined #ltsp
04:37
<polytan>
I think
04:37
I didn't put an option to use clear connection to save power
04:37
the "thin clients" are Core 2 duo
04:38
and the server a Mac Pro bi-quad core xeon
04:38
<knipwim>
sounds nice
04:41FreakGuard has joined #ltsp
04:42
<FreakGuard>
just tinkering a bit... does a user has to exist on a server or in the chroot?
04:44
<polytan>
on the server
04:45
<FreakGuard>
hm... can't login with the account on the server
04:45
<polytan>
a user in the chroot will only help you when going to a text shell with ctrl+alt+F1->6
04:46
<FreakGuard>
thanks for the info ;)
04:46
<Appiah>
polytan: what do you use to watch the videos?
04:47
you can try the different x outputs to see if they make any difference
04:47
<FreakGuard>
hmmm - does it matter that's not the ltsp-dm? ;)
04:47
<Appiah>
in mplayer you use -vo help to list all the different ones you can use
04:48synchris has quit IRC
04:54
<FreakGuard>
hmm... where is the ltsp dm started (debian) ?
04:55
<Appiah>
dm?
04:55
<knipwim>
display manager?
04:56
<polytan>
Appiah, I use VLC and totem
04:56
I don't really want to try with the security option of the connection desactivated
04:59
<Appiah>
I have no idea on how to force a specific video output for vlc
05:00
check in options if you can select xv,x11,gl,gl2,sdl etc
05:03Remaille has joined #ltsp
05:04
<knipwim>
http://www.chippc.com/thin-clients/jack-pc/
05:07
<FreakGuard>
knipwim, jep, display manager
05:08rjune has joined #ltsp
05:08
<FreakGuard>
I've got the problem that kdm starts... and that thingy doesn't really work as good as the default ltsp dm ;)
05:11
<knipwim>
FreakGuard: i guess the client
05:11
from the manual: The X server will either start an encrypted ssh tunnel to the server, in the case of ldm ...
05:12
tinkering myself as well
05:13Nubae has joined #ltsp
05:15
<FreakGuard>
oke ;)
05:19alkisg has quit IRC
05:24
<ogra>
FreakGuard, you cant use kdm on the client, that wont work, you can use XDMCP to connect to the servers KDM though
05:25
though note that you will lose nearly all functionallity, ltsp is built around LDM, all functions like sound, localapps, local devcies etc require the ssh tunnel for communication
05:32FreakGuard has quit IRC
05:52hanthana_ is now known as hanthana
06:03vvinet has joined #ltsp
06:12synchris_odium is now known as synchris
06:43phantom has quit IRC
06:56phantom has joined #ltsp
06:59alkisg has joined #ltsp
07:02hanthana has quit IRC
07:08Remaille has quit IRC
07:21RobertLaptop has quit IRC
07:22RobertLaptop has joined #ltsp
07:30RobertLaptop has quit IRC
07:31RobertLaptop has joined #ltsp
07:34FreakGuard has joined #ltsp
07:35
<FreakGuard>
how to set the displaymanger to ldm? doesn't seem to be in alternatives of the ltsp-chroot
07:38
<ogra>
its the default if you dont fiddle with the chroot :)
07:39evilx has joined #LTSP
07:39
<ogra>
LDM doesnt have an initscript, its integrated in the ltsp-client init process
07:39
so it isnt integrated with the alternatives system either
07:40
<FreakGuard>
well, I like tinkering
07:40
so just disable the DM that pops up?
07:41
<ogra>
uninstall it
07:41
i assume you use debian or ubuntu, right ?
07:41
<FreakGuard>
debian
07:41jammcq has joined #ltsp
07:42
<jammcq>
good morning #ltsp
07:42
<ogra>
yeah, just remove it, if it doesnt work, move your /opt/ltsp/i386 dir out of the way and re-run ltsp-build-client, that will work in any case
07:42
<FreakGuard>
that's not the point :P
07:42
<ogra>
(in case you did more tinkering that might interfere ;) )
07:42
<FreakGuard>
do I have to do update-image everytime I change the chroot?
07:43
<ogra>
not in debian
07:43
debian uses nfs by default
07:43
<FreakGuard>
hmm
07:43
<ogra>
ubuntu uses nbd, debian *can* use nbd but doesnt do so if you dont tell it to
07:44
so the image isnt used there
07:44
hey jammcq
07:44
<jammcq>
hey ogra
07:44
<ogra>
jammcq, have you seen http://www.linuxdevices.com/news/NS9634061300.html ?
07:45
<FreakGuard>
hm... what is better? ;)
07:45
I suppose nfs just sucks concerning performance, so...
07:46
<ogra>
nbd is faster
07:46
<Nubae>
in the end only ubuntu has nbd on by default?
07:46
<ogra>
but more maintenance work
07:46
since oyu need to rebuild the image after you made changes to the chroot
07:46
<FreakGuard>
well, that's just a bit of time
07:46
how to enable it? ;)
07:47
<Nubae>
farely fast on most chroots, but not so fast on fat chroots...
07:47
<ogra>
so for tinkering nfs is surely better, for production where you dont change anything, nbd is the better choice imho
07:47
<FreakGuard>
I'm only testing it here :D
07:47
some VM
07:47
<Nubae>
yah, there's the added benefit of the image being realiable cause it can't be messed up
07:47
<FreakGuard>
so basically I crash the whole system :D
07:48
I'm trying to use kde 4.x instead of 3.5.x
07:49
anyway
07:50hanthana has joined #ltsp
07:51
<ogra>
for kde4 you should probably use kubuntu ...
07:51
i think debian is still quite behind on the 4.x packages
07:52
<FreakGuard>
I don't really like debian, and ubuntu even less :/
07:53dmaran has joined #ltsp
07:53
<FreakGuard>
and debian has pretty new kde packages afaih...
07:54
<ogra>
well, the kde4 team works very closely with upstream in ubuntu
07:54
and for a DE thats still so young i would really look for the very latest to get some stability :)
07:56
<FreakGuard>
hm... where is the KDE 3.5 in the chroot?
07:56
aptitude doesn't show anything (?)
07:57
<ogra>
the chroot doesnt contain any userspace stuff
07:57
its only to bring up the clinet, start X enable the HW and run a DM
07:57
you then log in to the server to run a desktop session
07:57* FreakGuard fails horribly
07:59davidj has quit IRC
07:59phantom has quit IRC
07:59F-GT has joined #ltsp
08:00
<FreakGuard>
ogra, so the only interessting things are lts.conf and the ldm init scripts?
08:04alkisg has quit IRC
08:04
<ogra>
lts.conf actually
08:04
why would you do anything to the initscripts :)
08:05
<FreakGuard>
like change desktop?
08:05
hm... btw - where to change the mirror for ltsp-build-client?
08:06
I'm sitting almost next to a debian mirror
08:07
<ogra>
--mirror :)
08:07
ltsp-build-client --help
08:08
update-alternatives --config x-session-manager on the server lets you change the session
08:11
<FreakGuard>
thanks. it's still a bit confusing ;)
08:12
<ogra>
and ldm has a menu for it as well to select your session
08:14Gadi has quit IRC
08:15Gadi has joined #ltsp
08:15
<FreakGuard>
There is only 1 program which provides x-session-manager
08:15
(/usr/bin/startkde). Nothing to configure.
08:15
but why is KDE 3.5 started?
08:16
hmm
08:16
<-- fails
08:18
<ogra>
because you need to hack up your own startkde for kde4 in debian ? i think in kubuntu its a direct replacement
08:18
<FreakGuard>
hm... I might have a look at gentoo ;)
08:21pmatulis has joined #ltsp
08:24
<knipwim>
FreakGuard: i made some prelimenary documentation using gentoo's ltsp-5 port
08:24
not finished by a long shot
08:25
<FreakGuard>
hmm
08:25
<knipwim>
http://en.gentoo-wiki.com/wiki/LTSP
08:26
<FreakGuard>
doesn't debian have something like vimdiff for new config files?
08:26
<ogra>
you mean ucf ?
08:26
<FreakGuard>
This Vim was not compiled with the diff feature.
08:26
AAAAARGH
08:46din_os has joined #ltsp
08:55din_os has quit IRC
08:55
<johnny>
knipwim, syslinux should get pulled in automatically by ltsp-server
08:56
no need to manually install it
08:58
<knipwim>
i know
08:58
but then i have to install ltsp-server first in the manual
09:05mikkel has quit IRC
09:09din_os has joined #ltsp
09:13alkisg has joined #ltsp
09:20mikkel has joined #ltsp
09:21PerlStalker has joined #ltsp
09:32cliebow has joined #ltsp
09:35DarKnesS_WolF has joined #ltsp
09:35
<johnny>
knipwim, that's fine?
09:35
also.. the path for the tftp serfver should be /var/lib/tftpboot
09:35
or rather /var/lib/tftpboot/${arch}
09:35alkisg has quit IRC
09:37
<knipwim>
what's fine?
09:37
<johnny>
what's the big deal if you instal ltsp-server first
09:37
in fact.. i recommend it
09:38
also.. just do the layman thing first
09:38
so you can just do layman -a ltsp
09:39
this is unnecessar to edit thequickstart file
09:39
stage_uri="file:///opt/ltsp/stage3-x86-current.tar.bz2"
09:39
just do --stage_uri=/path/to/stage on ltsp-build-client
09:39
<knipwim>
check
09:40
i thought ltsp was only accessible with git
09:40
<johnny>
yes.. layman can require git
09:40
so emerge layman with the git use flag
09:40
and btw.. unionfs.. is not necessary
09:40
yet..
09:41
not sure why you're mentioning unionfs at all
09:41
then again.. if you don't want this to be part of the official instructions.. that's fine
09:41
but it works fine wihtout unionfs
09:41
we do not want to require patched sources at all
09:41
<knipwim>
i realized later
09:42
initially i used donnie berkholz's setup.txt
09:42
where he used unionfs
09:42
read about it, and thought it was a good solution
09:43
<johnny>
that was written awhile.. before unionfs stopped being able to build as external module
09:43FreakGuard has quit IRC
09:43
<knipwim>
i'll make a mention that union mounts are not required and can be skipped
09:50CAN-o-SPAM_ has joined #ltsp
10:02
<johnny>
knipwim, are you familiar with guide xml?
10:07
<knipwim>
not really, but i can learn :)
10:08
<johnny>
that's the official gentoo doc format..
10:08
it's very simple
10:08
i appreciate your effort up to this point btw.. i've been meaning to do that :)
10:08
just been very busy
10:08
<knipwim>
thanks
10:09
until now it has been a big learning experience
10:10
<PerlStalker>
Can anyone tell my why I'm getting "nbd_server[12620]: Read failed: Inappropriate ioctl for device" when a thin client starts?
10:11
<johnny>
and it doesn't continue on?
10:11
<knipwim>
i'm sure it will
10:12
<johnny>
that was to PerlStalker
10:12
<PerlStalker>
johnny: No. The client dies.
10:12
<johnny>
so.. if you find perl, are you gonna kill it?
10:13
<PerlStalker>
No. I just like secretly following it around and peeking through it's windows at night.
10:13
<stgraber>
Ryan52, sbalneav: We're investigating a bug with ldm + x11vnc which makes X to crash when you're logging in while a client is connect to the VNC server
10:13
<johnny>
hmm... where are the nbd people who might know this
10:14
stgraber, any ideas on this nbd issue?
10:14
<stgraber>
it looks like it's triggered when the greeter disappears
10:14
and not by any of the rc.d scripts
10:14
any idea ?
10:14
johnny: what issues ?
10:14
<johnny>
read up about 20 lines
10:15
<stgraber>
PerlStalker: is that with 8.10 ?
10:15cliebow has quit IRC
10:15
<stgraber>
PerlStalker: there is a race condition in the nbd code that's fixed in Jaunty
10:16
<PerlStalker>
Lovely.
10:16
<Ryan52>
stgraber: x crash or ldm/gtkgreet crash? 0.o
10:16
<PerlStalker>
I wonder why it was working for me before. *shrug*
10:17
<stgraber>
Ryan52: well, one of them :) the result being X and ldm to be respawned
10:17
PerlStalker: that's why we call that race condition :)
10:17
<PerlStalker>
Indeed.
10:18
<Ryan52>
stgraber: okay, probably ldm/gtkgreet then :)
10:18Remaille has joined #ltsp
10:18
<stgraber>
Ryan52: does ldm send SIGKILL to X when it dies ?
10:19
<Ryan52>
no, but we run x like "xinit ldm". (or something like that)
10:19
so when the child process dies, x exits.
10:20
what I usually do is just start x manually, then in the terminal that it opens I run ldm.
10:32
<stgraber>
Ryan52: "Exit request sent", that's what I get right after I loose everything
10:32
*before
10:35vvinet has quit IRC
10:44ajohnson has quit IRC
10:45
<Gadi>
stgraber: you get that in ldm.log?
10:48vvinet has joined #ltsp
10:48
<stgraber>
Gadi: nope, stdout
10:48
Gadi: I get two "exited with status 0" in ldm.log
10:49
hmm, hang on a sec, it's not the right bug :) in trying to reproduce it we did a mistake :)
10:49
<Gadi>
that sounds like x11vnc talking, then
10:49
<stgraber>
no, we've all process separate
10:50
this one is clearly ldm or ldmgtkgreet
10:50
<Gadi>
right, but "Exit request sent" is not a ldm/gtkgreet error message
10:53
<stgraber>
well, it's returned by ldm, I have no idea where it comes from though :)
10:53
as it's not in ldm's code
10:53hanthana has quit IRC
10:54
<stgraber>
Ryan52: how critical is Gtk-CRITICAL ?
11:01aledec has quit IRC
11:03vvinet has quit IRC
11:06staffencasa has joined #ltsp
11:07epsas has joined #ltsp
11:07
<epsas>
Hello
11:08
<stgraber>
Gadi, Ryan52: ok, just did some tests, it's clearly when ldm exits when VNC is opened that it makes X to crash
11:08
<epsas>
I am sure someone has run into this challenge before..
11:08
<stgraber>
when testing with other softwares, it's just fine
11:08
and when testing without the VNC connected, X doesn't reset (started with -noreset)
11:08
<epsas>
I want users to be able to customize their desktop
11:09
but I want their Desktop directory to include the contents from another directory (shared by other users)... on a read-only basis
11:09
ie, i want their Desktop/ to overlay another read-only directory
11:10
so we can have common commands and files available to users -- but still allow them to populate and customize their desktop
11:10Lns has joined #ltsp
11:11
<epsas>
Oh... and has there been any work on getting enabling an LTSP lab to function as a grid computing resource?
11:12
I know that there was an attempt at MOSIX earlier -- I do not know the progress of that
11:17Nubae has quit IRC
11:19Nubae has joined #ltsp
11:21
<din_os>
epsas: so you need some sort of common desktop icons such as the "all users" folder in windows
11:29
<Gadi>
stgraber: X always exits when ldm does
11:29
it is called by xinit
11:29
did you mean when the greeter exits?
11:35alkisg has joined #ltsp
11:37aglio2 has joined #ltsp
11:40spectra has joined #ltsp
11:40
<Lns>
din_os: an 'all users' equiv. homefolder/symlink/etc would be so nice to have in gnome/xdg setups in general
11:41
<din_os>
Lns, epsas: so it doesn't exist? if epsas responded I wanted to say I was interested in it too
11:41
<Lns>
seems it would be trivial to 'hack' together w/ a script to distribute to individual dirs and then in /etc/skel (like with "Examples" dir) but a more elegant solution would be cool
11:41aledec has joined #ltsp
11:41
<din_os>
I have managed to do it with scripts but... you know :) not the same
11:41
<Lns>
that's way out of the scope of ltsp though, that's an xdg thing really i'd think
11:42
exactly
11:42* Lns wonders about $xdg_data_dirs
11:43
<johnny>
Lns, how would you handle the permissions issues?
11:43
<Lns>
johnny: symlinks
11:44
<johnny>
huh?
11:44
symlinks doesn't solve permissions issues
11:45
hmm.. public
11:45
<Lns>
johnny: if you symlink to the real .desktop files then all you need to set perms on is the actual .desktop files
11:45
<johnny>
isn't that closed enough?
11:45danielsan474 has joined #ltsp
11:45
<johnny>
so.. is what you want an aggregation of all public directories?
11:45
AND.. just a folder for everybody?
11:46
<Lns>
johnny: is "Public" a shared thing? I thought it was something like "XYZ's shared documents" in win
11:46
<johnny>
yes
11:46
i assume you might as well do both at the same time
11:47
<Lns>
yeah
11:47
<epsas>
din_os - Yes. But I would like the contents of the "All users" folder on their desktop
11:47
a filesystem overlay
11:47
<Gadi>
have you tried using pam_mount +aufs?
11:48
seems to fit your bill exactly
11:48
<epsas>
I am also looking at http://en.wikipedia.org/wiki/UnionFS
11:48alkisg has quit IRC
11:48alkisg1 has joined #ltsp
11:49
<epsas>
it would be great to have desktop folders customized by group membership as well
11:49
<Gadi>
tell pam_mount to mount an aufs union of some shared read-only dir and say, $HOME/.desktop mounted to: $HOME/Desktop
11:49
<epsas>
so accounting would have their accounting apps, IS their tools, clients etc... etc..
11:50
<Gadi>
also easy with pam_mount
11:50
<epsas>
Gadi - got it. i am taking notes. =)
11:50
<Gadi>
since it is aware of groups
11:50
<Lns>
Gadi: don't you think that's a bit overkill? =)
11:50
<Gadi>
overkill?
11:50
<Lns>
epsas: have you looked at edubuntu-menus ?
11:50
<epsas>
Lns - the complexity can be abstracted by enough inspired scripting
11:50
<Gadi>
why?
11:50
<din_os>
epsas: in windows, if you put anything under c:\D&S\all users\desktop it appears on every user like you want it. But if you noticed the conversation here, there seems to be no similar folder in linux. Only script solutions
11:50
<Lns>
Gadi: well, a filesystem overlay just for common desktop icons
11:50
<epsas>
Lns - I am using Fedora
11:51
<Gadi>
thats what aufs is made for
11:51
<Lns>
Gadi: oh..i'll have to look at aufs :)
11:51
<epsas>
din_os - it looks like Gadi has non-scripting solutions
11:51
<Lns>
epsas: well it's just some script deals i think. i'm sure its portable with a little effort
11:51
<ogra>
epsas, edubuntu-menus uses xdg, no reason why the setup shouldnt work on fedora
11:52
<Gadi>
xdg solves menus, for sure
11:52
not launchers on the desktop, tho
11:52
<johnny>
that's for sabayon
11:52
which would be nice if someboyd would pick it up and make it work again :)
11:52
i can't afford to spend time on it
11:52* Lns doesn't like the menu / panel / desktop launcher separation
11:53* johnny == broke
11:53
<johnny>
relatively speaking at least
11:55danielsan474 has left #ltsp
11:56
<din_os>
epsas: let us know if you worked it out similar to windows. Post it here even if we are away, I'll check the logs
11:57aledec has quit IRC
11:58
<Lns>
holy crap ... http://packages.ubuntu.com/hardy/xdg-utils
11:59
<Gadi>
true. you could always use that and replace icons on login if they are removed
12:01pimpministerp has quit IRC
12:09polytan has quit IRC
12:10hanthana has joined #ltsp
12:16staffencasa has quit IRC
12:19RobertLaptop has quit IRC
12:19
<aglio2>
sbalneav: there?
12:22mikkel has quit IRC
12:24
<stgraber>
Gadi: I started ldm without xinit and it crashes my X when connected using VNC and doesn't when not
12:24
Gadi: running any other software (ldm-dialog or other) doesn't crash my X
12:30staffencasa has joined #ltsp
12:43alkisg1 has quit IRC
12:50aledec has joined #ltsp
12:57nothingman has joined #ltsp
12:58RobertLaptop has joined #ltsp
13:00alkisg has joined #ltsp
13:00ajohnson has joined #ltsp
13:02dewmon689 has joined #ltsp
13:02
<_UsUrPeR_>
ogra: ping
13:03Gadi has left #ltsp
13:05vagrantc has joined #ltsp
13:13vvinet has joined #ltsp
13:14
<dewmon689>
hey
13:16din_os has quit IRC
13:17aledec has quit IRC
13:17aledec has joined #ltsp
13:18CAN-o-SPAM_ has quit IRC
13:21Gadi has joined #ltsp
13:35bobby_C has joined #ltsp
13:35japerry has joined #ltsp
13:38CAN-o-SPAM_ has joined #ltsp
13:38CAN-o-SPAM_ is now known as CAN-o-SPAM
13:39
<CAN-o-SPAM>
Over 50 logo submissions for the new LTSP logo ... Review the submissions here: http://www.disklessworkstations.com/cgi-bin/web/contestlogos.html
13:39* vagrantc keeps forgetting to pay attention to the logo contest
13:40* CAN-o-SPAM has vagrantc covered
13:40
<vagrantc>
i keep thinking i'll read the emails... but they just sit there
13:41
CAN-o-SPAM: since there are multiple submissions from some people, it might be good to number them or something, so we don't get confused about which is which
13:41
<CAN-o-SPAM>
that should be done by the end of the week, they will all have some unique identifier
13:41
<vagrantc>
ok
13:41
<Lns>
CAN-o-SPAM: uh-oh, all those logos on one page might crash some thin clients... ;)
13:41
strangely ironic
13:41* vagrantc layghs
13:42* CAN-o-SPAM doesn't worry about crashing his Thin Client because his Atom chipset is so SWEET
13:42
<CAN-o-SPAM>
heh
13:42
<johnny>
i like lns's
13:43
with all the little penguins
13:43
<vagrantc>
there's some new ones since last i looked
13:43
<Lns>
lol
13:43
<vagrantc>
johnny: that's my favorite too, although it's a bit too busy for a logo
13:44
<Lns>
i'm really hoping that those penguin images aren't copyrighted..i didn't make them :( and i've scoured the web for hours looking for it's origination, the closest i could come up with was the quicktime 3 development project logo, whihc was a variation of that penguin
13:44
<vagrantc>
but it captures the networked server/client relationship, is cute and memorable, and looks clean ...
13:44
<Lns>
http://www.wap.org/journal/qtjan98/bigpenguin.jpg
13:45
I just don't want to get in trouble :)
13:45
<vagrantc>
i like the one with the feather, although it reminds me of apache...
13:46
<Lns>
personally, looking at the other logos especially, it's a bit much for a "logo"
13:46
vagrantc: ha, i thought the same thing
13:48Gadi has quit IRC
13:48
<vagrantc>
i also like the one by joshua higgins with the whispy things ... but it doesn't really have any substance to it
13:49* Lns 's 1220pxe is getting choppy from scrolling through logos
13:49
<vagrantc>
i really like the idea of demonstrating networked connectivity ... which is hard to do in a simple logo
13:49
<Lns>
i like this one - http://www.disklessworkstations.com/web/images/logos/aJabogado3.jpg
13:50
higgins..hehe..reminds me of a family guy episode.
13:51
<vagrantc>
i like thi middle one: http://www.disklessworkstations.com/web/images/logos/hVenugopalan123.png
13:55nothingman has quit IRC
13:56
<Lns>
ooo that one is nice too
13:56
<vagrantc>
seems like some people submitted multiple logos in a single image
13:56Gadi has joined #ltsp
13:57
<jammcq>
a logo needs to be simple. Imagine it being embroidered on a shirt. also think about what it would look like as a 16x16 pixel favicon
13:57
a logo rarely looks like what it is supposed to represent
13:58* Lns agrees with jammcq .. needs to be very simple
13:58nothingman_ has joined #ltsp
13:58
<jammcq>
classic examples are IBM, Coke, McDonalds
13:58
Debian
13:58
<Lns>
and of course, LNS.. ;) http://logicalnetworking.net/media/logo.jpg
13:58
<jammcq>
there's a logo that tells us little about what it is
13:59
<vagrantc>
debian's official logo is stupid and nobody wants to use it... everybody uses the unofficial logo :)
14:00
<Lns>
what's the debain logo(s) supposed to mean? Spiral, spiral coming out of a genie-bottle?
14:00
<jammcq>
FISL: http://programa.fisl.org.br/
14:00
<vagrantc>
if the logo can have a hair of representation, while remaining simple, that's great, though
14:00
<jammcq>
vagrantc: what's the unofficial?
14:01
<vagrantc>
jammcq: http://www.debian.org/logos/
14:01* jammcq thinks it's impossible to show anything representing a thin client in something as small as a logo. I think we should come up with a simple logo, and then come up with a nice graphic to be printed on T-Shirts if you want to show a Server/Thinclient setup
14:02
<knipwim>
i thought the penguins were nice, as in representing server-clients
14:02
<vagrantc>
jammcq: true enough
14:02
<jammcq>
knipwim: sure, for a big graphic to go on a T-shirt or website, but you can't really reduce that down to fit on a logo
14:03
<knipwim>
if penguins were eggs :)
14:03
<jammcq>
and donuts were dollars :)
14:04* vagrantc notes that many donuts cost more than a dollar
14:04
<jammcq>
Apache uses a feather. Tells us nothing about what Apache is, but it's recognizable as the Apache logo
14:04
<Gadi>
and just like a thin client, its not P.C.
14:04* vagrantc suspects it's tied to the whole apache indians and stupid cowboy movies
14:04
<Gadi>
:)
14:05
<jammcq>
postgresql uses an elephant. I suppose by some stretch, you could say that an elephant has a great memory, so it must be a great logo for a database
14:05
<vagrantc>
feather on the brow sort of thing
14:05
now we're blurring the line between mascots and logo
14:05* Gadi votes for Cupid
14:05
<jammcq>
mysql uses a dolphin. does a dolphin indicate a database?
14:05
<vagrantc>
they're clever and fast!
14:05
<Gadi>
he rides on top of the "cloud" and makes everyone fall in love
14:05
<Lns>
maybe all the animal mascots/logos are simply following tux
14:06
<epsas>
are muslims comfortable with all of the animal icons used by software projects?
14:06
i was always curious
14:06
<jammcq>
dunno, good question
14:06
<Lns>
why would muslims not like animal logos?
14:06
<vagrantc>
epsas: as a whole, i suspect you'll find a variety of opinions on the matter
14:07
<epsas>
Lns - because of the whole "Hold no false idols" thingy that Moses was scolded about
14:07* Gadi does not think muslims pray to OSS
14:07
<epsas>
but plants and geometrical patterns are fine
14:07jhutchins_wk has quit IRC
14:07
<vagrantc>
oh, that old thing
14:08
had no idea plants were given the go-ahead :)
14:08
<Lns>
epsas: ah..well that bleeds over into lots of other religions as well
14:08
my mom is a hardcore christian and she never liked me watching He-Man - she thought he was a false idol
14:08
<epsas>
Lns - that's true...
14:09* Gadi wonders what she thinks of American Idol
14:09
<jammcq>
heh
14:09
<Lns>
LOL
14:09
<jammcq>
or Billy Idol
14:09
<Lns>
quite interesting, she loves dancing with the stars
14:09
should bring that up to her =p
14:09
and she seemed to idolize McCain before the election
14:10
<jammcq>
sorry, was Billy Idol too old for this crowd?
14:10
<Lns>
jammcq: billy idol is a legend!
14:10
<vagrantc>
jammcq: i'm looking for a white wedding
14:10* Gadi thinks mccain's too old for this crowd
14:10
<jammcq>
vagrantc: nice day for it, eh?
14:11
<vagrantc>
jammcq: nice day to start again
14:12
<Lns>
i loved his cameo in Wedding Singer
14:12* jammcq was just thinking about that too
14:12
<jammcq>
cuz in 1st class, they pretty much let passengers do what ever they want
14:12dmaran has left #ltsp
14:14
<Lns>
hahaha
14:14
<vagrantc>
anyone who has a clue why ldm 2.0.36 hangs on logout, but ldm 2.0.34 works fine will make my day...
14:15
<jammcq>
even vagrantc is quoting movie lines now
14:15* Lns doesn't know how to translate 5.0.40~bzr20080212-0ubuntu4 to the actual version #
14:16
<vagrantc>
Lns: the version in bzr *-trunk on 20090212 , with 4 ubuntu-specific revisions
14:16
<Lns>
ah
14:16
well i'm way behind here on hardy then =p
14:21
<alkisg>
vagrantc, does it hang on unsuccessful login? (wrong password?)
14:24
<vagrantc>
alkisg: no, after sucessful login, on logout.
14:24
<alkisg>
Hmmm... tough one...
14:24
<vagrantc>
which seems very similar to some race conditions that were introduced and fixed last july
14:25
when we did the code re-write ...
14:25
the hard part is, there's no changes in the code i can see between working and broken versions.
14:25
that would seem to affect it
14:27
<alkisg>
Well, the hard way would be to use bisection... take a revision somewhere inbetween, compile it, try it etc...
14:29
For 32 revisions, it would take 5 compilations maximum
14:30
<vagrantc>
1106 to 1122
14:30
<alkisg>
...even less :)
14:41dewmon689 has quit IRC
14:42din_os has joined #ltsp
14:46vvinet has quit IRC
14:48vvinet has joined #ltsp
14:49spectra has quit IRC
14:57SpEcLeDx2 has joined #ltsp
15:03din_os has left #ltsp
15:09
<SpEcLeDx2>
hello everyone I am having an issue with flash audio while using nxclient to connect to an LTSP 5 server, when I play any flash with audio I get this error on the command line...ALSA lib pcm_pulse.c:625:(pulse_prepare) PulseAudio: Unable to create stream: Invalid argument....Could not set hardware parameters
15:12
<johnny>
it's probably the nx part
15:12
not ltsp
15:12
did you try it with plain ltsp?
15:14
<SpEcLeDx2>
it is
15:14
I have narrowed it down to that
15:15
everything works fine when I just use a normal LTSP connection
15:15
the problem is that we are trying to use nx to relieve a problem with xulrunner for firefox which causes major menu lag.
15:16
we also need to use it for remote users
15:16
the latter being the main reason
15:16
<johnny>
there's a known issue
15:16
not sure if it is fixable
15:16
<SpEcLeDx2>
I have been beating my head against a wall with this for the past two weeks
15:16
<johnny>
without sever hackery
15:17
<SpEcLeDx2>
which the firefox or sound issue?
15:17
<Lns>
SpEcLeDx2: does PA sound work otherwise under nx?
15:17
<johnny>
oh good quesiton
15:17
<SpEcLeDx2>
not remote however if I boot a thin client using nx it does
15:18
I have no sound remote at all
15:18
<Lns>
boot a thin client using nx?
15:19
as in, non-ltsp session?
15:19
<SpEcLeDx2>
i.e. I am working from home today and I have no audio when using the nxclient on my notebook however if I boot our thin clients in the office using nx I have audio in everything other then flash
15:19
<Lns>
weird
15:19
SpEcLeDx2: what version of flash/ltsp/os ?
15:19
<SpEcLeDx2>
its more along the lines of a limited x session
15:20
flash 10 ltsp 5 Ubuntu 8.10
15:20
<Lns>
not that it helps now but you should try using your laptop at your office
15:20
<SpEcLeDx2>
NX only supports ESD which I am not sure but could be causing the problem
15:20
<Lns>
ah that's probably it
15:20
<SpEcLeDx2>
I do no audio
15:20
<Lns>
you need to pipe esd through pulse if it isn't
15:21
Gadi would probably be able to tell you all about that
15:21
or ogra
15:21
<SpEcLeDx2>
that is what I have having a hard time doing sorry I am all over the place here but its been crazy,
15:22
wouldn't it be the other way around though? since anything using esd seems to play but when an app wants to use pulse nada?
15:23
<Lns>
well as you said, nx doesn't support pulse
15:23
i dont know if there's a way to reverse that
15:23
pulse is kind of the eden of pulling all other audio subsystems together and making them play nice
15:24
<SpEcLeDx2>
i c
15:24
<Lns>
imho anyway ;)
15:24
it's also fairly new compared to esd/oss
15:24
so integration into other apps/systems is trickling in, but not universal
15:24
<SpEcLeDx2>
there isn't a way to get flash/firefox to use esd is there?
15:26
<Lns>
SpEcLeDx2: iirc there's an option in /etc/firefox prefs or something to tell it which audio system to use..that was necessary a while back to get it to play with ltsp..don't remember what though, try googling
15:28
<SpEcLeDx2>
did and its gone in FF3 at least from what I have found/not found
15:28
:D
15:28
fun freakin fun
15:28
lol
15:28
<Lns>
bah
15:28
hehe
15:28
<SpEcLeDx2>
I am grasping at straws now
15:29
<Lns>
i could kill the dev team for going to sqlite (and not even considering alternative sql backends for multi-user setups like ltsp)... what a bunch of crap..but oh well =)
15:29
<johnny>
huh? alternative sql backends? lol
15:29
that requires servers
15:29
a browser should not require a server..
15:30
<Lns>
johnny: i mean more of a plugin architecture so you can choose.
15:30
<johnny>
lol
15:30
i don't think it would help
15:30
is the problem nfs mounted home?
15:30
<Lns>
it would help me.. just think about shared bookmarks for one example
15:30
<johnny>
wrong way
15:30
use an extension
15:30
<SpEcLeDx2>
i am pretty unhappy with a lot of the changes they made as well at least from a business stand point
15:30
<Lns>
johnny: what extension?
15:30
<johnny>
and a local deliciously
15:30
err server
15:31
deliciouslike*
15:31
like in moodle or something
15:31
if you're into that
15:31
didn't say it existed
15:31
just that you're talking about the problem from the wrong side
15:31
<SpEcLeDx2>
skuttle is a good one
15:31
<Lns>
not when FF2.x had bookmarks.html which was easily shared
15:31
and they took away the functionality of that
15:31
<johnny>
that was still the wrong way to do it
15:31
<SpEcLeDx2>
scuttle even
15:31
<Lns>
but it wokred...
15:31
worked*
15:32
<johnny>
doesn't make it right
15:32* Lns slaps johnny with a large trout
15:32
<johnny>
evolutionary progress works like this
15:32
some solutiosn get dead ended
15:32
and we get stuck with pulse before it was ready
15:32
and hal for that matter
15:32
and udev
15:32
etc
15:33
we're just getting on a path of stabilization in the past year or so
15:33
now it's tryign to make the apps place nice
15:33
<Lns>
personally i don't agree with that
15:33
<johnny>
altho imo ubuntu for example. i don't think it was a good choice
15:33
to back alsa with pulse
15:34
<Lns>
*nix philosophy has always been about being able to do anything and have it not break, even when new things come out
15:34
<johnny>
it's never been like that
15:34
that was a fantasy
15:34
except in corporate caclified environments
15:34
<Lns>
at least linux philosophy
15:34
<johnny>
no
15:34
it has never been like that
15:34
lots of drivesr have disappeared
15:34
for example
15:35
because there was nobody to maintain them anymore
15:35
linux is about picking up the slack if nobody wants something you want
15:35
otherwise it dies
15:36
<Lns>
bah i say, bah
15:36
<johnny>
you would be living a pipe dream then
15:36
i haven't been involve for all that long
15:36
only 7 years
15:36
but that's all i've seen
15:36
stuff that used to work.. stops working.. all the itme
15:37
as i mentioned earlier, it's only in the past year that stabilization has started
15:37
still more kinks to work out i'm sure.. and breakages ahead.. but the rate of change is slowing
15:39
<Lns>
SpEcLeDx2: scuttle...do you use it? how does it work exactly?
15:40
<johnny>
Lns, they went with sqlite.. because they know that all this eventually is just for local storage of remote sources
15:40
so.. sqlite just happens to be the chosen cache
15:40
<Lns>
looks like it just puts your bookmarks in a web-based db.. i'd like something that integrates directly w/the bookmark system
15:40
<johnny>
you're looking for something like googlegears
15:40
also integrated into html5..
15:40
<Lns>
johnny: I guess, but it sure doesn't do much for multi-user setups where you want to share information
15:40
<johnny>
but i forget what it's called
15:40
Lns, that's because they haven't been built yet
15:41
we're still working on it.. don't you worry
15:41
and once again.. reiterating what i said.. people get left in the lurch all the time
15:41
pulseaudio and hal are big examples of this
15:41
pain happens with evolutionary growth
15:42
i spent HOURS mucking with pulse on the computer at the store
15:42
it was terrible
15:42
during hardy
15:42* Lns wants to hire someone to create a mysql-based firefox extension that integrates into bookmarks so people can share bookmarks
15:42
<johnny>
no..
15:42
lns.. wrong way wrong way
15:42
wrong wray
15:42
<Lns>
right way for me! =)
15:42
<johnny>
no
15:43
what you want.. is a mysql basaed website that stores bookmarks
15:43
as a web service
15:43
<Lns>
i don't want a website for bookmarks
15:43
<johnny>
that the local boomark manager can talk to
15:43
<Lns>
i want the bookmarks in firefox
15:43
<johnny>
they will be cached locally for offline usage
15:43
you can serve the site locally
15:43
<Lns>
well either way
15:43
<johnny>
just like you'd be required to serve the mysql db locally
15:43
no difference
15:43bobby_C has quit IRC
15:43
<johnny>
except one way will actually be used by millions (the way i mentioned)
15:43
<Lns>
why have a website when you can just talk directly to mysql?
15:44
<johnny>
because REST is the future
15:44
this is just how ti's going to be
15:44
<Lns>
huh?
15:44
<johnny>
whether you like it or not :)
15:44
i'm telling you a way that will save you all ths money on extension development :) because people are already making what i said
15:44
<SpEcLeDx2>
Lns: yes it does put them in a database/website works really well we use it at work for our IT group
15:45
<johnny>
SpEcLeDx2, but does it integrate with the bookmarks manager in firefox?
15:45
<SpEcLeDx2>
there is a plugin for ff that allows you to add it to your scuttle server
15:46
extension whatever you call them
15:47
<Lns>
SpEcLeDx2: ^^^ what johnny said
15:47
<johnny>
Lns, people are going to make waht i said.. if it's not already existing
15:47
that is a guarantee
15:47
<Lns>
johnny: looking into REST
15:47
haven't heard of it
15:47
<johnny>
REST is just a technique
15:48
talking in simple HTTP verbs
15:48
GET,PUT,DELETE
15:48
POST
15:48
Lns, that's how webdav works
15:48
those verbs too
15:48
<Lns>
k.... /me is still confused
15:49
<johnny>
people won't be coding SELECT
15:49
<Lns>
how do you easily integrate this stuff with things like bookmark mgr
15:49
existing software
15:49
<johnny>
firefox extensions?
15:49
<Lns>
with their own way of doing things
15:49
<johnny>
existing software like what?
15:49
<Lns>
k
15:49
you're just preaching a different way to accomplish the same goal..you need an extension
15:50
<johnny>
right now... more and more of it will be hitting firefox.. the general support for local storage
15:50
like googlegars
15:50
google gears* is already hitting firefox
15:50Remaille has quit IRC
15:50
<johnny>
it's part of html5
15:51
SpEcLeDx2, the interesting thing to me.. is reusing already existing elements that people are used to
15:51
like the bookmarks manager
15:51
but storeing it on a remote server
15:52
<SpEcLeDx2>
ic
15:52
I think you lost me
15:52
:-)
15:52
<johnny>
uhmm.. you already know how to use the boomarks manager right???
15:53
<SpEcLeDx2>
yes
15:53
lol
15:53
<johnny>
imagine if you could sync the whole thing (or just selected tags) to a place that is shared by others
15:53
<SpEcLeDx2>
gotcha
15:53
<johnny>
no bookmarklets
15:53
<SpEcLeDx2>
now that makes sense and is a good idea
15:53
<johnny>
it's going to happen :)
15:53
<Lns>
johnny: your enthusiasm scares me =p
15:54
<johnny>
firefox3.1 will make it possible.. as long as they freakin fix their tracemonkey bugs :)
15:54
<SpEcLeDx2>
are you predicting the future here or do you know something we don't?
15:54
<johnny>
scares you? :)
15:54
if i know what you don't.. it's only because i am a web developer first
15:54
and i keep up on browser tech
15:55
<SpEcLeDx2>
makes sense then
15:55
<johnny>
i'm sure you know a whole lot i don't in some other area
15:55
<SpEcLeDx2>
what other exciting new technologies can we expect in the near future?
15:55
<johnny>
huge advances in accessibility
15:56
<Lns>
johnny: seems like google gears focuses on online vs offline access to data
15:56
<johnny>
the end of the death grip IE6 has held over the web
15:56
no.. google gears is the offline part
15:56
the online part .. is what you're already used to..
15:56
gmail,reader, etc
15:56
ie6 was the president bush of the web..
15:57
that's the analogy i've taken to using lately :)
15:57
if it weren't for ie6... flash wouldn't have became "the standard" that it is today
15:57* Lns points his pitchfork at IE6
15:58
<johnny>
the techniques you're going to be seeing on the web in the next year or so.. have been available for ages
15:58
but they were unusable in ie6
15:58
even something as simple as having fixed position boxes was impossible
15:58
without js
16:00
<vagrantc>
alkisg: so, by bisect, do you mean pick a revision in the middle, then the middle of whichever was broken, and so on?
16:00
<alkisg>
Yeah, like binary search
16:00
git has git-bisect for this
16:00
I had to do it some days ago for gpxe :(
16:01
<vagrantc>
what would you do at the revision control level?
16:01
<alkisg>
Erm, rephrase please?
16:01
<vagrantc>
alkisg: well, talkinga bout git-bisect got me confused
16:02
<alkisg>
Like you said it: 1106 to 1122 => the middle is 1114, you try this one first
16:02
If it fails, you take the middle of 1104 to 1114...
16:03
<vagrantc>
since everything between a known working version and 1112 are all copyright updates, maybe i'll just start with that
16:03nothingman_ has quit IRC
16:04
<alkisg>
git-bisect automates that, it just asks you "did it work or not?" and handles itself the revision bisection. But I don't know if it can work with bzr branches, and anyway it shouldn't need more than 4-5 compilations, so it's easy enough to do it manually...
16:05
<vagrantc>
i should also just try and roll a tarball myself, i've been using ubuntu's .orig.tar.gz ...
16:12jammcq has quit IRC
16:30
<yanu>
vagrantc: only gnome-session isn't working on the clients, with LDM_REMOTECMD i can start kde
16:30
<vagrantc>
well, this is *really* interesting. with an ldm 2.0.36 tarball i rolled myself, it seems to be working just fine.
16:31
<yanu>
vagrantc: so far the moment users cannot choose there DM/WM in LDM
16:31
<vagrantc>
yanu: i think ldminfod from ldm-trunk fixes your issue...
16:32
yanu: http://bazaar.launchpad.net/~ltsp-upstream/ltsp/ldm-trunk/annotate/head%3A/ldminfod/ldminfod
16:33
or: http://bazaar.launchpad.net/%7Eltsp-upstream/ltsp/ldm-trunk/download/head%3A/ldminfod-20081031022502-ehq85efyvrlwsedi-3/ldminfod
16:37
<yanu>
oh, i have a workaround, so the kids can do there thing :)
16:37
vagrantc: that works indeed, now they can choose
16:38
<vagrantc>
yanu: you could also set the default with: update-alternatives --config x-session-manager
16:38
<yanu>
vagrantc: still gnome-session refuses to work, but i'll live with that
16:39
<vagrantc>
yanu: right, but if your default is broken, i'd recommend figuring that out or setting a different default :)
16:39
<yanu>
that last one i did, the 'figuring out'-one is a bit more difficult
16:40
<vagrantc>
sure, but simply setting a different default on the server will prevent you from having to mess with lts.conf
16:40
<yanu>
thanks for the hint 'update-alternatives'
16:40
<vagrantc>
LDM_REMOTECMD is rather hackish
16:41
<yanu>
i uncommented it
16:41
<vagrantc>
uncommented?
16:41
<yanu>
put a # in front of it -> # LDM_SESSION=
16:42
<vagrantc>
uncommented would mean you removed the # in front of it :)
16:42
<yanu>
indeed, how do you call that?
16:42
<vagrantc>
commented
16:43
<yanu>
ach ok, haha, something wrong in my head
16:43
<vagrantc>
un at the beginning of a word usually means the opposite :)
16:44
<yanu>
hmm, the opposite, i always have problems with that, left-rigth, ok-nok , ... dyslexy perhaps :)
16:44* ogra checks who pinged him
16:44
<ogra>
ha
16:44
err
16:44
ah
16:45
Lns, pulse installs the esound compatibility module by default in ubuntu
16:45
so it should just work
16:45
<Lns>
ogra: ah
16:45
<ogra>
i think in debian too, though i'm not 100% sure here
16:45
Jan 1 04:22:23 in-target: en_AU.UTF-8...
16:45
Jan 1 06:10:11 in-target: done
16:45* ogra glares at the timestamps
16:45
<Lns>
yeah..since hardy, i haven't had to touch sound on ltsp...
16:45
<vagrantc>
it used to work, but it's been a while since i tested the esound stuff
16:46
<ogra>
speedy arches are lovely, arent they ? :)
16:46
<vagrantc>
only 50 minutes to generate a single locale?
16:46
<ogra>
yeah
16:46
<vagrantc>
ouch.
16:46
<ogra>
NSLU2 install
16:46
133MHz, 30MB
16:46
<Lns>
oh gross..i used to have one of those =p
16:47
<yanu>
ogra: try the speeding up tric 133 -> 266MHz
16:47
<ogra>
and our langpack-en contains *all* en_XX.UTF-8 locales :P
16:47
<Lns>
horrible, horrible default firmware
16:47
<yanu>
ogra: just solder a resistant out
16:47
it works great
16:47
<ogra>
yanu, no, i'm developing the ubuntu port on it, i need to use it unmodified :)
16:47
<yanu>
ok then
16:47
<ogra>
(indeed i need to fix that issue)
16:47
<Lns>
why would you want to put ubuntu on that??
16:48
this is linksys NSLU2 right?
16:48
<ogra>
to use it as a NAS
16:48
<yanu>
because 'they' can ?
16:48
<ogra>
because its ARM hw that has a community and we want omeone to help testing our ARM port ;)
16:48litlebuda has joined #ltsp
16:49
<vagrantc>
well, since stgraber's ldm 2.0.36 tarball isn't working for my and Ryan52 has at least one important fix, i may tag 2.0.37 soonish...
16:49* ogra would have preferred the beaglebard but that will have to wait for 9.10 if we get the kernel merged
16:49
<yanu>
mine is running debian, but i know someone who doesn't use his nslu2 ... time to mail him again
16:51
<Lns>
ugh..i can't believe those things are even still being sold
16:51
(sorry..just a real bad experience w/linksys NAS devices)
16:52
<yanu>
Lns: i use it as a nightly backupsystem with ninjabackup, also it serves a webserver, and isrri is always running with bitlbee (irc, msn, jabber, gmail, ...)
16:53
i can't mis it
16:55
<ogra>
i know some people using it as mediaserver in their LAN
16:55
works great
16:56
and then there is the soldering iron league, there are apparently some insane people that soldered 128M on it etc
16:56
<vagrantc>
package updates sound brutal, though
16:56
<ogra>
ell, depends ... if you attach a fast USB HDD and use lots of swap it will be ok
16:57
getting d-i to not hit OOM all the time was the big challenge
16:58
<Lns>
you guys think the linksys hardware itself is good enough to trust? I know their firmware is absolute crap now a days, but you have these communities w/the wap54g, and others that seem to think that the hardware is top notch
16:58
<ogra>
the ubuntu kernel wasnt really designed for 30M
16:58
<vagrantc>
heh. yeah, considering the install mounts a tmpfs with something like 40-60MB ... ?
16:59
<stgraber>
vagrantc, Gadi: Does that pulse change I just pushed make sense ?
16:59
it really improves performances on video using sound when running as localapp
16:59
<vagrantc>
stgraber: pushed where?
16:59
<stgraber>
ltsp-trunk
16:59
<vagrantc>
ah good. no idea!
16:59
but i'm tagging ldm-trunk now :)
16:59
<ogra>
yeah, well, i got it working now partman uses swap as soon as its available, so just getting through the partitioning is enough
17:00
<vagrantc>
stgraber: for some reason, your ldm 2.0.36 tarball didn't work for me. but when i rolled my own tarball, it was fine...
17:00
<stgraber>
vagrantc: hmm, I'll likely retag then as there's something in ldm that crashes when x11vnc is attached to it
17:01
that's really a problem with iTalc as when the teacher as italc opened, nobody can login as ldm keeps making X to crash
17:01
<vagrantc>
ouch
17:01
<stgraber>
haven't found what it's yet but it's clearly something that ldm or ldmgtkgreet does right before exiting
17:02
as it doesn't happen with any other X application
17:02
and happens if you enter a correct or incorrect login
17:02
<vagrantc>
ldm-trunk 2.0.37 tagged and pushing...
17:02* ogra mumbles ... en_DK.UTF-8 ... why do danes have their own english locale
17:02
<stgraber>
(note: that's with X -noreset)
17:03
<vagrantc>
stgraber: i'm guessing you're using a newer autotools or soemthing that mkdst is calling...
17:03alkisg has quit IRC
17:03
<vagrantc>
drove me crazy... couldn't find any code changes in bzr ... rolling my own tarball works fine.
17:03
<stgraber>
vagrantc: yeah, might be an autotools issue
17:04
<vagrantc>
*crazy*
17:04
<ogra>
do we use libtool ?
17:04* vagrantc trembles with craziness
17:04
<ogra>
i know jaunty had a new libtool that changed a lot of things
17:05
<stgraber>
ogra: all the previous packages were made from Intrepid
17:05
<ogra>
ugh
17:05
<stgraber>
ogra: I only updated my netbook to Jaunty yesterday :) (was using my lappy for Jaunty testing but not for LTSP)
17:06
<ogra>
https://lists.ubuntu.com/archives/ubuntu-devel/2008-November/026862.html
17:07
<stgraber>
ogra: right but an upstream tarball is supposed to work for all distros no ? or we should ship a generated configure in our tarballs
17:08
<ogra>
thats what mkdst is for
17:08
fix mkdst
17:08
debian sid uses the same libtool btw
17:08
at least it should now after lenny is out
17:10
<vagrantc>
still 1.5.26-4 for stable/testing/unstable
17:10
2.2.6a-1 in experimental
17:10
<ogra>
ah
17:11
2.2.6a-1ubuntu1
17:11
in jaunty
17:11
but sid should switch soon
17:11
and i think fedora uses 2.x.x as well
17:12
<vagrantc>
right now, i want to get ltsp-related packages moving to testing so i can do proper backports for lenny :)
17:13
<stgraber>
vagrantc: didn't push the tag ? I see the commit but not the tag
17:13
<vagrantc>
oh hell.
17:14
stgraber: i tagged it as 2.0.37 ... not ldm-2.0.37 ...
17:15
<stgraber>
ah, right :)
17:15
<vagrantc>
sorry about that
17:16
if anyone happens to have any commits, please tag it properly and push...
17:17CAN-o-SPAM has quit IRC
17:18
<stgraber>
should be fixed, if bzr works as I hope it to work :)
17:19
basically a non-change commit and the fixed tags
17:19
<vagrantc>
in my experience, it's never pushed tags unless there was a "real" commit to push, and it never pushes deletions
17:20
but i'll try it
17:21
<stgraber>
well, I pushed that change, no idea if it worked though
17:22
<vagrantc>
seems to have
17:22
<ogra>
i see two tags for the former commit in olive
17:23artista_frustrad has quit IRC
17:23
<stgraber>
so the tag wasn't removed ?
17:23
<ogra>
and no tag for the no-change commit
17:24
well, olive shows two tags, not sure how accurate that is ... its only a gui tool
17:24
but my branch wasnt up to date and thats what i got with my recent 20 versions i pulled
17:24artista_frustrad has joined #ltsp
17:25
<vagrantc>
i don't think tag deletions can be pushed
17:25
short of deleting the remote branch and re-pushing
17:26vvinet has quit IRC
17:27
<ogra>
hmm, i also see both tags in the tag scrollbar in olive
17:29
why does ldminfo.c say "/* LTSP Graphical GTK Greeter" in the copyrigt note ?
17:31
<vagrantc>
because i cut and pasted from where the code originally came
17:31
i./e. the code was split out into a separate file at some point ...
17:31
alternately, i cut and pasted from the wrong location
17:37
uploading ldm 2.0.37-1 to debian unstable ...
17:37* ogra grins ...
17:38
<ogra>
i have actually never read the development policy doc
17:38
<vagrantc>
heh
17:38
i think sbalneav and i hashed it out a while back
17:38
<ogra>
i dont break peoples thumbs !
17:38
<vagrantc>
heh
17:38
<ogra>
i show up at their door !
17:39
<vagrantc>
ogra: don't encourage bad code!
17:39
<ogra>
so i should break thumbs ?
17:39* ogra finds that a bit harsh for curly brackets
17:39
<vagrantc>
it's only an idle threat
17:40
so maybe codifying violence in our codingstyle document is a bit out of line
17:40
<johnny>
lol
17:40
<ogra>
a bit
17:40
<johnny>
that is funny
17:42
<Gadi>
stgraber: ping
17:43
<vagrantc>
ogra: feel free to update the codingstyle document :)
17:44
<stgraber>
Gadi: pong
17:45
<Gadi>
stgraber: did you ever fix that x11vnc problem?
17:46
<stgraber>
Gadi: nope, it's really weird
17:46
Gadi: and a major bug when used with iTalc
17:46
as basically you can't login when the teacher is monitoring the classroom
17:46
<Gadi>
any messages in ldm.log, like: Sending greeter the quit command failed
17:46
or some such?
17:46
<vagrantc>
stgraber: i'd worry it wouldn't know to use pulse at all without PULSE_SERVER set ...
17:47
<stgraber>
I have an easy way of reproducing it though
17:47
vagrantc: we set .asoundrc for that and alsa's pulse plugin will look for the unix socket
17:47
Gadi: I just installed: xnest, x11vnc and ldm on my laptop
17:48
then started Xnest, start x11vnc on it then started ldm and opened a vnc connection
17:48
then logged in from that vnc
17:48
and boom, you get a X segfault
17:48hanthana_ has joined #ltsp
17:48
<stgraber>
but no segfault or anything similar on ldm's side ...
17:48
<vagrantc>
stgraber: i know the pulse/alsa layers are a bit different for debian and ubuntu
17:48
so i'd just worry a bit about compatibility
17:49
<Gadi>
does X actually segfault or just exit?
17:49
<stgraber>
Program received signal SIGSEGV, Segmentation fault.
17:49
http://paste.ubuntu.com/123065/
17:49
<ogra>
https://bugs.launchpad.net/ubuntu/+bug/312168 ?
17:51
<stgraber>
nope
17:51
I have that on Intrepid and Jaunty
17:51
only happens with ldm, if I start it once the user is logged in, I can use it for hours with 30 computers without a crash
17:51
<ogra>
hmm, there is an update request
17:51
<stgraber>
and it only happens when ldmgtkgreet exits
17:51
<ogra>
https://bugs.launchpad.net/ubuntu/+source/libvncserver/+bug/312994
17:52
you should probably test if the new upstream fixes it
17:52
<stgraber>
happens with iTalc as well which has a snapshot of upstream (yeah, that's bad :))
17:53
<Gadi>
stgraber: can you pastebot ldm.log?
17:54
<ltsppbot>
"stgraber" pasted "ldm.log" (26 lines) at http://pastebot.ltsp.org/265
17:54PerlStalker has quit IRC
17:54
<stgraber>
note that it crashes X with or without a valid login or server
17:54
so for my tests I don't use a valid login and don't set a server
17:55
<Gadi>
test@None
17:55
thats interesting
17:56
ah - you dont set up a proper env for ldm, it seems
17:56
might want to at least export LDM_SERVER
17:57
<stgraber>
as I said, as it doesn't make a difference whether or not the environment is right, I just use an empty one :)
17:58
<Gadi>
can you trace the processes, so you know which ids go with what?
17:58
oh, nm
17:58
its in the log
17:58
hehe
17:59johnny has left #ltsp
18:01
<Gadi>
stgraber: does autologin cause a segfault, too?
18:02
<stgraber>
yes
18:03
hmm, actually I'm not sure. Guest login does
18:03
complete autologin is harder as you'd need to connect the VNC before ldm exits to test :)
18:06hanthana has quit IRC
18:06
<Gadi>
you can run the greeter independently of ldm, right?
18:06
I think
18:06
ie. launch xterm within Xnest
18:07
and just run the greeter itself
18:07
without ldm
18:07
and see if it causes a segfault
18:07
<stgraber>
Gadi: right, but can't make it to exit :)
18:07
<Gadi>
not even when you enter username/password?
18:08gentgeen__ has quit IRC
18:08
<Gadi>
I thought it exits by itself
18:08
<stgraber>
nope, I believe it's trying to talk to ldm
18:08
<Gadi>
run a window manager then, flip to the xterm and kill it
18:08
:)
18:09
<stgraber>
killing it doesn't trigger the segfault, I tried it :)
18:09
<Gadi>
great minds...
18:09litlebuda has quit IRC
18:10
<Gadi>
can you replace it with a script that just dumps the right stuff to stdout?
18:11
<vagrantc>
just a simple little script...
18:12
<stgraber>
hmm, just changed the greeter to exit when I click on reboot and it doesn't trigger the segfault ...
18:19johnny has joined #ltsp
18:19vvinet has joined #ltsp
18:20
<stgraber>
ldmgtkgreet exiting doesn't seem to be the trigger as my reboot button hack (calling gtk_main_exit) doesn't make it crash but entering a login/pass does
18:20
<Gadi>
even without ldm?
18:21
<stgraber>
yes, ldmgtkgreet standalone, exits just fine without any crash
18:21
<Gadi>
unless you enter un/pw
18:22
<stgraber>
ldm+ldmgtkgreet when exiting before trying to login works just fine too, the same after the login sequence (so login, fails => exit 1) triggers the segfault
18:22
<Gadi>
can you try calling close_greeter()
18:22
instead of gtk_main_exit()
18:24
<stgraber>
where is close_greeter() defined ?
18:24
<Gadi>
greeter_iface.c
18:25Egyptian[Home] has quit IRC
18:25
<stgraber>
undefined reference
18:25
that's a ldm function
18:26
<Gadi>
http://lists.freedesktop.org/archives/xorg/2009-February/043715.html
18:26
hmm... interesting....
18:27SpEcLeDx2 has quit IRC
18:31
<Ryan52>
stgraber: did you get it all figured out?
18:33
<stgraber>
not really no .. I suspect something's wrong in close_greeter
18:33
<Ryan52>
so wait, when does it crash?
18:33
right after you connect to vnc?
18:33
<stgraber>
no, when the greeter disappears and I'm connect on vnc
18:34
and that's an X crash, not ldm ...
18:34
but likely triggerd by ldm somehow
18:34
so for now I commented everything in close_greeter. So the greeter doesn't get killed.
18:34
ldm exits correctly and the greeter is still there
18:34
then I manually kill the greeter
18:34
and everything is fine, no crash
18:35* Ryan52 is still lost
18:35
<Ryan52>
so it's a seg fault?
18:35
or what?
18:35
<stgraber>
yes, a X segfault
18:35
<Gadi>
Ryan52: http://paste.ubuntu.com/123065/
18:35
<stgraber>
that only happens when your run ldm and x11vnc and is only triggered when ldmgtkgreet is closed by ldm
18:35
<Ryan52>
so then it's not our fault :)
18:36
<Gadi>
it seems that close_greeter must be too brutal, so whatever pointer VNC is expecting to use disappears
18:36
<stgraber>
well, I never saw it happening outside of ldm and can reproduce it every-time with ldm so ...
18:37* stgraber is trying to cut in that function to see what makes it crash
18:38
<Ryan52>
all it does is tell ldm to quit, tho.
18:38
<vagrantc>
Ryan52: you calling LDM a quitter?
18:38
<Ryan52>
:P
18:39
and I meant gtkgreet anyway :P
18:39
<Gadi>
stgraber: try commenting out the cursor change
18:39
in greeter.c
18:39
<stgraber>
found it
18:39
not the fix, but the problem :)
18:40
<Gadi>
do tell
18:40
<ltsppbot>
"stgraber" pasted "hack" (26 lines) at http://pastebot.ltsp.org/266
18:40
<Gadi>
right
18:41
<stgraber>
no more crash after that part commented
18:41
<Gadi>
now try unhacking that
18:41
and hack gtkgreet/greeter.c
18:41
<Ryan52>
that extra newline really helps, eh? :P
18:41
<Gadi>
line 186
18:41
comment out 186-188
18:42
I think the issue is with:
18:42
186 GdkCursor *cursor;
18:42
187 cursor = gdk_cursor_new(GDK_WATCH);
18:42
188 gdk_window_set_cursor(gdk_get_default_root_window(), cursor);
18:43
<vagrantc>
Ryan52: after adding an ldm-server binary to ldm, debian/rules is failing at # ensure ldm-script is executable to avoid lintian warning chmod +x debian/ldm/usr/share/ldm/ldm-script
18:43
<stgraber>
Gadi: no more crash
18:43
<Ryan52>
vagrantc: uhhh...what?
18:43
<Gadi>
by commenting out the cursor?
18:43
<Ryan52>
did you push it?
18:43
<stgraber>
Gadi: yeah
18:43
<vagrantc>
Ryan52: not yet
18:44nothingman_ has joined #ltsp
18:44
<Gadi>
yeah, so it seems it doesnt like changing to the watch icon
18:44
:)
18:44
<Ryan52>
vagrantc: good, cause I hate bzr. then can you pastebin the patch?
18:44
<Gadi>
maybe Ryan52 knows
18:44
he's the resident gdk guy
18:44
<Ryan52>
who? what? huh?
18:45
<Gadi>
Ryan52: this causes the segfault:
18:45
186 GdkCursor *cursor;
18:45
187 cursor = gdk_cursor_new(GDK_WATCH);
18:45
188 gdk_window_set_cursor(gdk_get_default_root_window(), cursor);
18:45
<Ryan52>
hmm. that causes X to crash?
18:45
<Gadi>
yeah
18:45
<Ryan52>
that *really* shouldn't happen afaik.
18:45
<stgraber>
Gadi: funny part is that when opening a gnome session, the same cursor works fine :)
18:45
<Gadi>
maybe the pointer to the root window gets mucked up with x11vnc
18:45
<Ryan52>
are you sure that it's not just causing ldm to crash?
18:46
<ltsppbot>
"vagrantc" pasted "fails to build after adding ldm-server package..." (80 lines) at http://pastebot.ltsp.org/267
18:46
<stgraber>
Ryan52: I ran ldm in gdb, no crash ...
18:46
<vagrantc>
Ryan52: ^^
18:46
<Ryan52>
stgraber: so if you run ldm from an xterm, does it still crash? or are you just left with your xterm?
18:46
<stgraber>
Ryan52: you can easily reproduce the bug if you want, you just need a x11vnc connected on your X server and a client looking at the screen , then login, then boom
18:46
<ogra>
you might get a different rootwin
18:46* vagrantc wonders if Ryan52 has multi-threading implemented
18:47
<Ryan52>
ogra: still don't see how that would cause X to crash, tho.
18:47
<ogra>
no, not really
18:47
<Gadi>
well, whats the negative of not setting the cursor there?
18:47staffencasa has quit IRC
18:47
<stgraber>
Ryan52: I'm opening it on a Xnest for testing and my X is still very much alive after ldm exit, so works just fine
18:47
<Gadi>
just that it stays normal until the gnome session picks it up?
18:47
<ogra>
Gadi, it looks ugly and you dont get a busy cursor
18:47
<stgraber>
Ryan52: previous behavior was ldm exiting correctly, X segfaulting, x11vnc exiting and the client too
18:48
<Ryan52>
Gadi: we don't see the nice little watch symbol thingy.
18:48
<Gadi>
ogra: well, you get it when the session starts, no?
18:48
<Ryan52>
oh, ogra answered already.
18:48
<ogra>
Gadi, thats an important usability feature to show its checking the pw
18:48
<Ryan52>
Gadi: no, between gtkgreet exiting and session starting, which can take a while.
18:48
<ogra>
right
18:48
<stgraber>
ogra: that's not in the check password part, that's in the exit function
18:48
<Gadi>
ah
18:48
<Ryan52>
ogra: that's not in that part of the code.
18:48
<ogra>
oh
18:49
<Gadi>
yeah, what they said
18:49
<Ryan52>
oh, and stgraber already said *that*.
18:49
<ogra>
hmm, the last action you did was checking the pw though
18:49
<Ryan52>
vagrantc: it's not implemented very well :P
18:49
<ogra>
so just dont change the cursor back and you wont need to set the clock cursor again
18:49
<vagrantc>
Ryan52: debian/rules ?
18:50
<Ryan52>
vagrantc: 16:47 * vagrantc wonders if Ryan52 has multi-threading implemented
18:50
<vagrantc>
ah. :)
18:50
<ogra>
heh
18:50
<vagrantc>
Ryan52: i'm wondering if it's the .PHONY: binary-indep
18:50
or is there a binary-all target ...
18:51
<Ryan52>
what? huh?
18:51
confused, one sec.
18:51
oh, hmm. are we mixing arch: any and arch: all?
18:51
that target still needs to be .PHONY.
18:51
but ya.
18:51
copy this:
18:51
binary-arch: build install dh $@
18:51
(with the \n\t that irrsi took out)
18:51
and change -arch to -indep
18:52
and it will work.
18:52
ohhhhhhhhhhhhhh.
18:52
<vagrantc>
Ryan52: ldm is arch any, ldm-server is arch all
18:52
<Ryan52>
okay, we haven't gotten to that part yet.
18:52
<vagrantc>
it's building it twice or something...
18:52
<Ryan52>
but do what I just said.
18:52
there's another change needed.
18:52
so can you just commit, push, and then I'll clean it up?
18:53
vagrantc: change debian/ldm/usr/share/ldm/ldm-script to debian/tmp/usr/share/ldm/ldm-script
18:53
and then add the binary-indep target.
18:53
and then you're all good.
18:54
<vagrantc>
Ryan52: change it to what?
18:54
<Ryan52>
vagrantc: which part?
18:54
<vagrantc>
Ryan52: the ldm-script ... ?
18:54
<Gadi>
stgraber: what happens if you move line 390
18:54
<Ryan52>
change:
18:54
chmod +x debian/ldm/usr/share/ldm/ldm-script
18:55
chmod +x debian/tmp/usr/share/ldm/ldm-script
18:55
<Gadi>
up to the global part
18:55
<Ryan52>
to that.
18:55
<vagrantc>
Ryan52: ah, i couldn't see the diff side-by-side :)
18:55
<Gadi>
and use "root" instead of the function to get the root window
18:55
<Ryan52>
hmm.
18:56
vagrantc: is that patch sposed to be done?
18:56
<vagrantc>
Ryan52: which patch, done how?
18:56
<Ryan52>
ldm-server.install is sort of wrong.
18:56
and you're missing ldm.install..
18:56
<vagrantc>
don't need it anymore
18:57
<Ryan52>
how will it know what to install? 0.o
18:57
also, you should (but don't have to) change the install target in debian/rules to use a stamp like build does.
18:57
<ogra>
if you dont use a ,install file it will use whatever make install did
18:58
<vagrantc>
we should really just ship ldm-script instead of generating it ... it doesn't use libexec anymore
18:58
<ogra>
*.install
18:58
<Ryan52>
ogra: not when you have multiple binaries packages, right?
18:58* Ryan52 may be completely lost..
18:58
<ogra>
no, then you need .install files
18:58
<Ryan52>
exactly.
18:58
<ogra>
thats right
18:59
<Ryan52>
so what you said was irrelevant to the current discussion.
18:59
<vagrantc>
after switching to multiple binaries we need to add the .install file back in?
18:59
<Ryan52>
yep.
18:59
<vagrantc>
got it.
18:59try2free has joined #ltsp
18:59
<ltsppbot>
"Gadi" pasted "stgraber: would this work?" (28 lines) at http://pastebot.ltsp.org/268
18:59
<vagrantc>
although, we really just want to install debian/tmp/* ...
19:00
<Ryan52>
yes.
19:00
that's what it does.
19:00
that's why chmodding debian/ldm/something will fail.
19:00
<vagrantc>
with, or without an .install file?
19:00Lns has quit IRC
19:00
<Ryan52>
oh, I see what you mean.
19:00* vagrantc will see in a moment
19:00* Gadi has to run - night, all
19:01Gadi has left #ltsp
19:01* Ryan52 confused himself there for a minute :P
19:01
<vagrantc>
we have two packages, one of which installs everything in debian/tmp (previously debian/ldm) and one of which installs a single file from upstream sources.
19:01
(ldm-server)
19:01
<Ryan52>
ya.
19:01try2free has left #ltsp
19:01
<stgraber>
gadi's patch doesn't work
19:01
<Ryan52>
vagrantc: I think you just have to specify everything.
19:02
<stgraber>
though what's weird is that the cursor when ldm starts appears correctly and doesn't crash X
19:02
<vagrantc>
Ryan52: ldm.install: debian/tmp/*
19:02
<Ryan52>
vagrantc: then it'll install ldm-server's thing too.
19:02
<ogra>
rm it from rules
19:02
<vagrantc>
Ryan52: ldm-server's thing doesn't get installed there
19:02
<Ryan52>
vagrantc: it doesn't? hmm...I think it should...
19:02
but ya, that'll work then.
19:04
<nothingman_>
anyone here have experience with using an airport-attached disk for /home?
19:05
<ogra>
nothingman_, wow, you love corner cases, dont you :)
19:06
<Ryan52>
vagrantc: just put "*" in debian/ldm.install, tho. not debian/tmp/*.
19:06
(I think)
19:06
<vagrantc>
oh.
19:06
well, i'll let me debian/tmp/* finish building and see what happens
19:07
<nothingman_>
ogra: well, in lieu of regular NAS, or a PC with NAS linux installed
19:07
<ogra>
nothingman_, how many users does your ltsp server have ?
19:08
<nothingman_>
I figure it'll be fast with little overhead
19:08
<ogra>
i cant imagine a wlan attached /home can handle many
19:08
<nothingman_>
and I already have the airport in place, underused
19:08
ogra: 10-20
19:09
<ogra>
thats a lot of constant writing ...
19:09
dont forget things like firefox cache in /home
19:09
what does airport give you ? 54Mbit ?
19:10
i would expect that to be constantly saturated
19:10
at least if your users use firefox and openoffice
19:11
<Ryan52>
stgraber: so the library call that's causing the problem is gdk_window_set_cursor? if you only comment out that line it all works fine?
19:12
<ogra>
Ryan52, just drop the code that sets back the cursor after pm checking
19:12
and add code that sets a default cursor on ldm startup
19:13
<Ryan52>
why add code that sets a default cursor?
19:13
<ogra>
to make sure its actually re-set once Gadi implements his evil plan to restart ldm without restarting X :)
19:13
<Ryan52>
oh.
19:14
I did not know about Gadi's evil plan.
19:14
<stgraber>
Ryan52: fun, it's only gtk.gdk.Cursor(gtk.gdk.WATCH) that makes it crash
19:14
<ogra>
he didnt work out how to make that stuff secure yet :)
19:14
<Ryan52>
stgraber: ahah. that was my suspicion.
19:15* Ryan52 wonders what language that might be.
19:15
<vagrantc>
Ryan52: well, debian/tmp/* worked, although for some reason i can't figure, ldm-script isn';t executed
19:15
er, executable
19:15vvinet has quit IRC
19:15
<ogra>
hack it up in debian/rules ?
19:15
<Ryan52>
we already do :)
19:15
<ogra>
i think debhelper makes everything -x by default
19:15
<vagrantc>
try to
19:15
<Ryan52>
is dh_fixperms touching it?
19:16
<vagrantc>
ogra: not in /usr/share/*
19:16
<Ryan52>
"It removes execute permissions from any libraries, headers, perl modules, or desktop files that have it set."
19:16
<stgraber>
Ryan52: http://paste.ubuntu.com/123097/
19:16
<ogra>
well, tmp/ is before it gets copied to the actual binary package subdir
19:16
<Ryan52>
hmm.
19:16
<vagrantc>
Ryan52: the chmod runs after dh_fixperms
19:17
<ogra>
try replacing tmp with the packagename
19:17
<stgraber>
Ryan52: you can choose if you want your X to dies or not by a command line parameter :)
19:17
<ogra>
and move it a little later in rules
19:17
<Ryan52>
stgraber: eww.
19:17
vagrantc: no, it doesn't.
19:17
<vagrantc>
ogra: with the packagename it fails entirely.
19:17
<Ryan52>
vagrantc: fixperms runs runs in a different place, I'm fairly certain.
19:17
vagrantc: try in binary-arch
19:17
<vagrantc>
Ryan52: looking at my build logs, it pretty clearly runs after dh_fixperms
19:17
<ogra>
where do you call it in rules ?
19:17
<Ryan52>
using debian/ldm/
19:18
<ogra>
is the target pkg arch all or any ?
19:18
<vagrantc>
two targets, one any, one all
19:18
<Ryan52>
stgraber: does that *always* crash in any vnc session?
19:19
<ogra>
and ldm-server is all ?
19:19
<stgraber>
Ryan52: yeah, start X, connect x11vnc to it and open a vnc client. Then run the script with the crash parameter an X crashes
19:19
<ogra>
then the chmod should be in arch-indep
19:19
<stgraber>
Ryan52: I tried with Xnest, Xephyr and regular X
19:20
<Ryan52>
stgraber: and back to it not being our fault :)
19:20
<ogra>
and point to the binary package dir name
19:20
<vagrantc>
i should fix it upstream rather than hack around it in this script.
19:20
<ogra>
how do you fix a dh issue upstream ?
19:20
<Ryan52>
dh is the one doing it.
19:20
ya, what ogra said.
19:20
<vagrantc>
because it's working around a stupid bug upstream
19:20
<Ryan52>
no, I'm pretty sure that dh is changing it's permissions.
19:21
<ogra>
no, its dh making it unexecutable
19:21
<Ryan52>
wait, where are we installing ldm-script?
19:21
I think we might be installing it in the wrong place..
19:21
<vagrantc>
Ryan52: /usr/share/ldm/ldm-script
19:21
<Ryan52>
it should be in libexec, which is lib on debian, I think.
19:21
<vagrantc>
it's getting installed in the right place, and it even works there
19:21
<ogra>
it needs to be /usr/lib
19:21
then dh wont make it unexecutable
19:21
<vagrantc>
Ryan52: we deprecated the whole idea of putting it in libexec.
19:22
<Ryan52>
vagrantc: why?
19:22
<vagrantc>
Ryan52: why bother?
19:22
<ogra>
FHS ?
19:22
<Ryan52>
because that's the correct place...
19:22
<ogra>
lintian ?
19:22
<stgraber>
Ryan52: right, but I'll need a way to fix it :) Having all my thin clients to crash on login because the teacher has italc open is a bit annoying :)
19:22
<ogra>
what Ryan52 said
19:22
<vagrantc>
we've been over this at least 3-4 times, and flip-flopped on the issue every time.
19:22
<ogra>
executable scripts need to live in /sr/lib
19:22
*usr
19:22
<Ryan52>
/usr/libexec/ upstream, tho.
19:22
I think.
19:23
<ogra>
whatever
19:23
<Ryan52>
:)
19:23
<ogra>
in the package it needs to end up in /usr/lib
19:23
<Ryan52>
yes.
19:23
so we should move it upstream to libexec.
19:23
<ogra>
i dont get why lintian didnt moan loudly
19:23
<vagrantc>
because it's a non-issue.
19:23
<Ryan52>
and then dh7 passes the right stuff to ./configure to make it end up in /usr/lib
19:23
<ogra>
right
19:23
<Ryan52>
vagrantc: executable files shouldn't live there.
19:24
so lintian should complain.
19:24
<vagrantc>
it's an architecture-independent executable.
19:24
<Ryan52>
maybe I:, tho.l
19:24
<vagrantc>
it's getting rather pedantic.
19:24
<ogra>
executable scripts live in /usr/lib on debian based systems
19:24
<Ryan52>
iirc FHS says that too.
19:24
<vagrantc>
i can show you thousands of executable scripts in /usr/share
19:24
<ogra>
/usr/share is reserved for shared *data*
19:25
then file packaging bugs :)
19:25* Ryan52 googles for FHS
19:25
<ogra>
/usr/share is wrong
19:25
it contains xml, icons default configs etc
19:25
<vagrantc>
well, re-open this ancient thread.
19:26
it's certainly not a library.
19:26
which is what FHS says /usr/lib is for.
19:26
<Ryan52>
/usr/libexec is for binaries, tho.
19:26
<vagrantc>
though i guess it does meet internal binaries that are not
19:26
intended to be executed directly by users or shell scripts."
19:27
<Ryan52>
vagrantc: ls /usr/lib/git-core/
19:27
<ogra>
right, /usr/libexec is for binaries ....
19:27
<Ryan52>
there's an example of a package that does it right.
19:27
<ogra>
/usr/lib/packagename is the debian equivalent for /usr/libexec
19:28aglio2 has quit IRC
19:30
<Ryan52>
and /usr/lib/pbuilder/
19:31
so anyway...
19:31
<vagrantc>
well, if you look at the log history for ldm-trunk/share/ldm-script.in it was last year february that we had these debates
19:31
so, i look forward to debating them again next year. it'll be a fine tradition.
19:31
<Ryan52>
lol
19:32
<vagrantc>
it was moved to libexec in mid-january, and then we decided to deprecate libexec by mid february.
19:32
<ogra>
grin
19:32
<stgraber>
Ryan52: do you wan't to see even more buggy ? :)
19:33
<Ryan52>
stgraber: sure.
19:33
<stgraber>
http://pastebin.com/f7205b052
19:33
this one will give you 10s of that cursor, then crash X
19:33
<Ryan52>
o noez.
19:33
hmm.
19:33
are we not freeing it?
19:33
could that actually matter?
19:34
oh, how can we.
19:34
gtkgreet is dead.
19:34
stgraber: if you add the python equivalent of gdk_cursor_destroy does it work?
19:34
(after the sleep)
19:35
<ogra>
gah, he said the magic word that made me look at my desktop clock
19:35
<Ryan52>
heh
19:38
<stgraber>
the closest I can find is to del() the variable but that doesn't change anything
19:38
<Ryan52>
del is a python thing?
19:39
<ogra>
yeah, C would be free()
19:39
<Ryan52>
ah, do you have unref?
19:40
gdk_cursor_destroy is just an alias for gdk_cursor_unref
19:43
<stgraber>
nope, can't find unref
19:50nothingman_ has quit IRC
19:58japerry has quit IRC
19:59
<stgraber>
Ryan52: looks like it's really the fact that it's animated that makes X to crash ...
19:59
I tried with all the others and none are making it crash
20:12
Ryan52: would checking for X11VNC and changing the cursor if detected be ok with you ?
20:13
Ryan52: X11VNC sets a X atom so I can look for it and switch to a static cursor when found
20:13aglio2 has joined #ltsp
20:14
<Ryan52>
that's ugly.
20:14
but sure.
20:14japerry has joined #ltsp
20:15
<ogra>
not setting the cursor to the hourglass doesnt solve it ?
20:15
i mean, just keep the cursor thats set my the pm checking function
20:16
<Ryan52>
oh ya, back to that.
20:16
<ogra>
*by
20:17
<stgraber>
ogra: no, if the cursor is WATCH and ldm exits it'll crash X, the only way is to have a static cursor when ldm exits
20:18
<ogra>
sounds like a theme issue
20:20
<vagrantc>
Ryan52: did you say try the chmod in the binary-indep target?
20:20
<Ryan52>
no.
20:20
you were going to...weren't you?
20:21
at least you never asked me to, so..
20:21
<vagrantc>
well, i've been testing before committing
20:21cyberorg has quit IRC
20:22
<Ryan52>
so what do you want me to do?
20:22
<vagrantc>
Ryan52: i remebered there was something you suggested that i hadn't tried yet...
20:23
ah, you said binary-arch ...
20:23
<Ryan52>
move the file to libexec? :P
20:23
but ya, move it to binary-arch.
20:23
and change the debian/tmp to debian/ldm
20:24
<vagrantc>
Ryan52: before or after "dh $@" ?
20:24
Ryan52: debian/ldm doesn't work- that's what it was before.
20:24
debian/tmp works.
20:24
<Ryan52>
yes, I know.
20:24
but it will work in this place.
20:24
but hold on.
20:24
<vagrantc>
??
20:25
<Ryan52>
okay.
20:25
change it to:
20:25
dh --before dh_fixperms binary-arch
20:25
chmod +x debian/ldm/....
20:25
err.
20:25
<vagrantc>
ah, this fun ...
20:25
<Ryan52>
lemme try that again :P
20:25
dh --until dh_fixperms binary-arch
20:25
chmod +x debian/ldm/....
20:25
dh --after dh_fixperms binary-arch
20:25
there.
20:26
<vagrantc>
so we're skipping the dh_fixperms step...
20:26
<Ryan52>
no.
20:27
we're putting "chmod +x debian/ldm/...." in right after it.
20:27
if I said --before instead of --until, then we would be skipping it.
20:27
(and that's what I said at first by accident)
20:27
not taking out any of dh's stuff, tho.
20:27
<vagrantc>
ah, got it
20:29pmatulis has quit IRC
20:29
<vagrantc>
could also exclude ldm-script and call dh_fixperms ourselves
20:30
<Ryan52>
that too..
20:30
<vagrantc>
dh_fixperms --exclude ldm-script or some such
20:30
<Ryan52>
ya.
20:30
what about ownership?
20:30
isn't dh_fixperms the thing that makes everything owned by root?
20:31
<vagrantc>
maybe ...
20:31
<Ryan52>
so then chmod +x would be better.
20:31
<vagrantc>
fair enough
20:47
<Ryan52>
vagrantc: is it working?
20:48
<vagrantc>
Ryan52: yeah, works like a charm!
20:48
<Ryan52>
yay!
20:48
<vagrantc>
debhelper 7 rocks.
20:48
<Ryan52>
now we should switch it to libexec so that you don't need that hack!
20:48
:P
20:48
<vagrantc>
heh
20:48
well, you'll have to fix upstream for that
20:48
<Ryan52>
vagrantc: it's about to get cooler, too. have you seen joey's blog post?
20:48
<vagrantc>
i mean, you can hack around in debian/rules, but why not just fix it upstream
20:49
Ryan52: not recently...
20:49
<Ryan52>
http://kitenet.net/~joey/blog/entry/debhelper_dh_overrides/
20:50
no more needing to use --before and --after
20:51jammcq has joined #ltsp
20:53
<Ryan52>
vagrantc: speaking of Debian packaging, you need to become a DD so your signature counts so I can sign up for the NM process :P
20:55* Ryan52 only needs a DD to sign his gpg key to be able to sign up :(
21:00johnny has left #ltsp
21:01ball has joined #ltsp
21:04Gadi has joined #ltsp
21:05
<vagrantc>
Ryan52: heh
21:07* stgraber will need a code review
21:14
<vagrantc>
Ryan52: have you applied for debian-maintainer? oh, you still need a signature for that, i guess.
21:30ball has quit IRC
21:32artista_frustrad has quit IRC
21:33artista_frustrad has joined #ltsp
21:43
<stgraber>
Ryan52: http://pastebin.com/f412eb452
21:45
note: I'm a very very very bad C coder, that code works but it's probably bad :)
21:48
<Ryan52>
ok.
21:50
<stgraber>
is there something I should have cleaned up or some other bad things I did ? :)
21:54
<Ryan52>
does it compile with -Wall? I don't think it does (tho that's just by looking at it).
21:54
and I think all you need is to use gdk_atom_intern..
21:55
no, I take that back it should compile with -Wall
21:55
and don't pass 1, pass TRUE. makes things more readable.
21:58
<stgraber>
gdk_atom_intern doesn't check for the property on the root window
21:58
<Ryan52>
oh.
21:58
<stgraber>
or if it does I don't see where we tell it where to look (current window or root window)
21:59
<Ryan52>
ya, nevermind.
21:59
<stgraber>
C is really complicated on this one btw
21:59
in python + gdk it'd have been
21:59
screen = gtk.gdk.screen_get_default()
21:59
root = screen.get_root_window()
21:59
if root.property_get(gtk.gdk.atom_intern("X11VNC_TICKER")):
22:00
<Ryan52>
heh
22:05
<stgraber>
pushed with these few fixes.
22:05johnny has joined #ltsp
22:05CaScAdE^FarAway has joined #ltsp
22:20CaScAdE^1arAway has quit IRC
22:51Gadi has left #ltsp
23:04cyberorg has joined #ltsp
23:16tarbo has quit IRC
23:17tarbo has joined #ltsp
23:21alekibango has quit IRC
23:21alekibango has joined #ltsp
23:31_Shane has quit IRC
23:34pimpministerp has joined #ltsp
23:38alkisg has joined #ltsp
23:41Egyptian[Home] has joined #ltsp
23:45Egyptian[Home]1 has joined #ltsp
23:50pimpministerp has quit IRC