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


Channel log from 15 April 2008   (all times are UTC)

00:17abadger1999 has quit IRC
00:23abadger1999 has joined #ltsp
00:30asac_ has joined #ltsp
00:42asac has quit IRC
00:42asac_ is now known as asac
00:47indradg_ has joined #ltsp
00:51leio has quit IRC
01:16deavid has joined #ltsp
01:17saxonwold has joined #ltsp
01:24saxonwold has quit IRC
01:35viking-ice has quit IRC
02:11slipttees has quit IRC
02:12slipttees has joined #ltsp
02:32primary has joined #ltsp
02:46ogra has quit IRC
02:46praveer_cool has joined #ltsp
02:47ogra has joined #ltsp
02:54viking-ice has joined #ltsp
02:59primary has left #ltsp
03:14indradg_ has quit IRC
03:25indradg has joined #ltsp
03:33mikkel has joined #ltsp
04:19
<jonkke>
Hi
04:20
is there any way that normal user could disable some printers from his/her profile without being lpadmin?
04:33praveer_cool has quit IRC
04:37FLOLO has joined #ltsp
04:41slipttees has quit IRC
04:52viking-ice has quit IRC
04:54Pascal_1 has joined #ltsp
05:08viking-ice has joined #ltsp
05:11pdjbarber has joined #ltsp
06:16viking-ice has quit IRC
06:18Pascal_1 has joined #ltsp
06:20TelnetManta has quit IRC
06:25Q-FUNK has joined #ltsp
06:40viking-ice has joined #ltsp
06:44Pascal_1 has quit IRC
06:45
<tarzeau>
hello
06:45
what generates the xorg.conf for ltsp?
06:46
it detects "vga" instead of "nv" on a g4 imac here
06:48
<Q-FUNK>
ogra: seems that there's a lot of running around in circle with the sync request for -geode. could you chip in your support by testing on all the geode hardware you have and reassure everyone that the thing works?
07:03deavid has quit IRC
07:03TelnetManta has joined #ltsp
07:06Pascal_1 has joined #ltsp
07:07mhterres has joined #ltsp
07:11muh2000 has joined #ltsp
07:12Pascal_1 has quit IRC
07:14
<vagrantc>
tarzeau: linux distro and release?
07:14
<tarzeau>
vagrantc: debian live + debian lenny + ltsp?
07:19
<vagrantc>
whoah
07:19
:)
07:20
tarzeau: so, in a typical ltsp setup, the default xorg.conf generation would be handled by /usr/share/ltsp/configure-x.sh
07:20
tarzeau: but debian-live may be doing some mucking with that.
07:20Guaraldo has joined #ltsp
07:20
<vagrantc>
tarzeau: you can disable ltsp's configuration with CONFIGURE_X=False in lts.conf
07:21
<tarzeau>
vagrantc: well debian-live generates a working one, i'll test it ...
07:21
<vagrantc>
tarzeau: ltsp's configuration mostly relies on X.org's autodetection
07:21
<tarzeau>
when i run /usr/share/ltsp/configure-x.sh
07:21
manually i get "vga"
07:22
if i run X -configure i get:
07:22
vga too, which is wrong...
07:22
then the X autodetection sucks
07:23
<vagrantc>
tarzeau: well, 7.3 has yet to migrate to testing ...
07:23
<tarzeau>
true
07:24
<vagrantc>
without knowing for sure, i get the impression huge improvements were made with X.org 7.3
07:24
<tarzeau>
experimental even has a newer version
07:24
let me try if i can download and install it from sid, on this live cd
07:25cliebow has joined #ltsp
07:26
<tarzeau>
yep it works with the xorg 7.3 from sid
07:27
daduke_: we could just put the *.debs from bee:/var/cache/apt/archives/ into the config/chroot_local-packages/
07:27
<daduke_>
vagrantc in the house!
07:28daduke_ is now known as daduke
07:31jammcq has quit IRC
07:31
<daduke>
vagrantc: FYI, this is in order to provide the LTSP live CD also on ppc
07:35
<vagrantc>
cool
07:37
daduke: you mean a single CD that does both powerpc and i386 ?
07:37
<daduke>
vagrantc: well I haven't tried that yet, but I saw something about multiarch boot.... so far it's 2 images.
07:57deavid has joined #ltsp
07:59exodos has joined #ltsp
08:04cyberorg has quit IRC
08:10
<Q-FUNK>
ogra: ping :)
08:18
<ogra>
Q-FUNK, please elaborate on the libDDC stuff, a bit more detail there will convince slangasek/pitti i think :)
08:19
sorry, i'm extremely busy this week, i'll see if i can make a test, no promises though ...
08:22Gadi has joined #ltsp
08:23
<Q-FUNK>
ogra: libDDC is not into the driver yet. it will come in 2.9.0. right now, the code differences are minimal and easy to handle. it's mostly about transitioning people to -geode and making them change their static xorg.conf, if any.
08:24
ogra: well, we mostly need as many supporters of this transition to chip in and give their thumbs up.
08:26cliebow has quit IRC
08:26cyberorg has joined #ltsp
08:51mikkel has quit IRC
09:11
<daduke>
vagrantc: https://www.phys.ethz.ch/~daduke/imac_ltsp.jpg
09:22FLOLO has quit IRC
09:25elisboa has joined #ltsp
09:37elisboa has quit IRC
09:38pdjbarber has quit IRC
09:39FLOLO has joined #ltsp
09:48gentgeen__ has joined #ltsp
10:01elisboa has joined #ltsp
10:31staffencasa has joined #ltsp
10:43cliebow has joined #ltsp
10:49mccann has joined #ltsp
10:50FLOLO has quit IRC
11:13abadger1999 has quit IRC
11:14mhterres has quit IRC
11:28exodos has quit IRC
11:28spectra has joined #ltsp
11:31jonkke has quit IRC
11:47abadger1999 has joined #ltsp
11:54Bengoa has joined #ltsp
11:57deavid has quit IRC
11:58elisboa has quit IRC
11:58deavid has joined #ltsp
12:05otavio has quit IRC
12:16rcc has joined #ltsp
12:19abadger1999 has quit IRC
12:27
<lns>
daduke, that's a lot of daisy-chained power plugs there.. =p
12:29rcc has quit IRC
12:30rcc has joined #ltsp
12:31rcc is now known as godfather
12:32godfather is now known as GodFather
12:40abadger1999 has joined #ltsp
12:47otavio has joined #ltsp
12:48gonzaloaf has joined #ltsp
12:52Pascal_1 has joined #ltsp
12:59Q-FUNK has quit IRC
13:01GodFather has quit IRC
13:01Pascal_1 has quit IRC
13:06leio has joined #ltsp
13:23gonzaloaf_work has quit IRC
13:23gonzaloaf_work has joined #ltsp
13:34deavid has quit IRC
13:37deavid has joined #ltsp
14:01milesd_ has joined #ltsp
14:16
<vagrantc>
otavio: any comments regarding: http://lists.alioth.debian.org/pipermail/pkg-ltsp-devel/2008-April/001341.html ?
14:20K_O-Gnom has joined #ltsp
14:23viking-ice has quit IRC
14:28abadger1999 has quit IRC
14:33
<otavio>
vagrantc: sorry, can't open it now. Can you give me an overview of the problem?
14:34
<vagrantc>
otavio: just need feedback on the architecture issue, and specific what the procedure i'll need to follow for architecture removals...
14:34
<otavio>
vagrantc: ah
14:34
vagrantc: well, feel free to reduce the arch list
14:35
<vagrantc>
otavio: at the moment, i'm thinking just i386, amd64 and powerpc ...
14:36
<otavio>
vagrantc: i guess they're the most used ones. Ack from my side
14:36
<vagrantc>
otavio: if someone offers to test and help us support more, i'd gladly add additional ones.
14:37
otavio: and then at some point we'll need to mail ftp-masters to remove the obsolete arches from unstable and testing?
14:43TheJoel has joined #ltsp
14:44
<otavio>
vagrantc: testing can be done throught RM team but sid does need ftp-masters
14:44
vagrantc: so the best is to ask for it to be removed from sid as soon as possible
14:45mhterres has joined #ltsp
14:45
<TheJoel>
Can I interrupt everyone for just a minute? I have a question...
14:46
<vagrantc>
!question
14:46
<ltspbot>
vagrantc: "question" is if you have a question about ltsp, please go ahead and ask it, and people will respond if they can. please also mention the linux distro and release you're using. :)
14:46
<TheJoel>
I have attempted the LTSP install, and no pxe image was created. How do you manually create the image file?
14:46
<vagrantc>
otavio: although if RM can remove it from testing, then unstable -> testing migration will work just fine, yes?
14:47
<otavio>
i guess so
14:48
<vagrantc>
otavio: i mean, i'll try and communicate with all necessary parties. :)
14:48
TheJoel: which linux distro and release?
14:48
<TheJoel>
Ubuntu 7.10.
14:50cliebow has quit IRC
14:50
<vagrantc>
TheJoel: there isn't a PXE image, just a kernel and initrd.img and pxelinux.0 and configuration
14:50
TheJoel: should show up in /var/lib/tftpboot/ltsp/*
14:50milesd_ has quit IRC
14:51
<TheJoel>
So you don't need the "pxelinux.0"?
14:51
<johnny>
yes you do
14:52indradg is now known as wellfed
14:53Egyptian[Home] has quit IRC
14:53cpunches has quit IRC
14:53ivazquez has quit IRC
14:53moquist has quit IRC
14:53loather-work has quit IRC
14:53Paladine has quit IRC
14:53* lns starts a wave
14:53
<vagrantc>
otavio: thanks for your comments :)
14:53
<otavio>
vagrantc: np
14:53
<TheJoel>
So then, which online tutorial would you recommend for the LTSP setup?
14:54moquist has joined #ltsp
14:55
<johnny>
i just used ltsp-build-client
14:55
and then setup my dhcp server properly
14:55
that's all i had to do
14:55
since i already had a seperate dhcp server setup
14:56
i just had to integrate ltsp specifics
14:57wellfed is now known as indradg
15:01TelnetManta has quit IRC
15:06cpunches has joined #ltsp
15:06
<vagrantc>
TheJoel: hmmm... http://changelogs.ubuntu.com/changelogs/pool/main/l/ltsp/ltsp_5.0.40~bzr20080212-0ubuntu5/changelog
15:07
first entry
15:07Paladine has joined #ltsp
15:07
<johnny>
vagrantc, !
15:07
would you be averse to addine one more elseif in ldm for LDM_XSESSION var?
15:08Egyptian[Home] has joined #ltsp
15:08ivazquez has joined #ltsp
15:08
<vagrantc>
johnny: yes, i would rather get it fixed properly.
15:08
<johnny>
is it fair that we are the ones who are compromised ? and everybody else got their wish? :)
15:08
<vagrantc>
we should really just define it at build time.
15:09
<johnny>
i didn't quite understand the objection to warren's idea
15:09
<vagrantc>
fair, no, but is it fair that debian and ubuntu are at the bottom of the list even though we were using it first?
15:09
<warren>
which idea?
15:09
<johnny>
doesn't matter where you are on the list.. just that you are there :)
15:10
for letting the server figure out which session to run
15:10
<vagrantc>
johnny: it requires server-side stuff and i don't see any issue with a more simple fix.
15:10
<warren>
vagrantc: technically the server SHOULD be deciding what to run
15:10
<johnny>
well.. his way seems to allow more future growth with less effort
15:10
<vagrantc>
warren: agreed. and i think we should keep that in ldminfod
15:10
<warren>
johnny: that's my point exactly
15:12
<vagrantc>
well, nobody's said that in response to ogra's or my post.
15:12
<warren>
johnny: anyhow, both my idea and server telling the client the location of ldminfod require a protocol change of sorts
15:12
vagrantc: that was part of my original post on the subject
15:12
johnny: and I rather not change the protocol multiple times
15:13
johnny: I think we're just going to think of a way to eliminate ldm in the next 6 months in favor of something in gdm.
15:13
<johnny>
well.. that's why i wanted my XSESSION variable in there temporarily.. so i don't have to wait it out or manually patch it
15:13
i'd prefer my stuff to be on the same level as all other upstream
15:13
<vagrantc>
warren: your original post didn't really address my proposal made in response to your posting.
15:14
<johnny>
so we can figure it out.. but i'll still have something working in the meantime
15:14
<warren>
I dislike the idea of having fallbacks encoded in the client too.
15:14
<johnny>
sure.. but you're relying on it now
15:14
<warren>
Stuff on the client filesystem do not necessarily match the stuff on the server.
15:14
<vagrantc>
agreed, but it's a reasonable default.
15:14
<warren>
I disagree that it is a reasonable default
15:14
<vagrantc>
ok.
15:14
<warren>
You only don't want to change it from the way it was
15:15
It is a BAD DESIGN for the client to have to decide this particular thing
15:15
it should have been the server's decision from the start
15:16
Please understand, I am only speaking technically, I'm not emotionally tied to one way or another.
15:16
I just think we're better off ldm client picking a standard script path and the server choosing what to do.
15:16
That way ldm client in the future requires no real changes.
15:16
while the server could change
15:16
<vagrantc>
well, i agree that it should have been server-side originally. but i also think my proposal prefers the whatever is set server-side, and only has a fallback for when that doesn't work.
15:17
<warren>
"fallback for when that doens't work"
15:17
given that it runs a command through ssh
15:17
how does the client reliably know if it failed?
15:17
<vagrantc>
warren: so then define your default at build-time to be that value.
15:17
warren: i.e. if you want ldm to default to /etc/ltsp/Xsession or whatever, then set it as such.
15:17
<warren>
tftp is a fragile protocol
15:17muh2000 has quit IRC
15:18
<warren>
If lts.conf failed to pull down it might try to do the default defined at build-time that has nothing to do with the current network
15:18viking-ice has joined #ltsp
15:18
<vagrantc>
i'd prefer to keep it configurable rather than hard-coded.
15:18
<johnny>
i think warren's option is most configurable
15:18
<warren>
vagrantc: I might propose that we rip out the if/else hardcoded locations we have now and instead build-in your own distro's default.
15:19
<vagrantc>
johnny: it's explicitly less configurable.
15:19
johnny: my proposal includes everything warren proposed with an additional option to configure at build time.
15:19
warren: did you even read my proposal?
15:19
<warren>
which thread name?
15:20
<vagrantc>
ldm: ConsoleKit and Xsession location
15:20
posted on the 8th of april
15:21
<warren>
> i dont like #2, it would mean you cant just connect to any ssh server at
15:21
> > all.
15:21
i agree with oliver; i don't like #2.
15:21
<vagrantc>
but my proposal doesn't preclude a distro from implementing that option.
15:21
if you want that for your distro, then make that the default behavior.
15:22
<warren>
I don't mind keeping the LDM_SESSION option
15:22
however I think it is highly misguided to actually use it
15:22
then have an rc.d script (or code within ldm after the ssh tunnel is up)
15:22
that validates the LDM_SESSION values in order, using the first one that
15:22
is valid.
15:22K_O-Gnom has quit IRC
15:22
<warren>
vagrantc: isn't this a bit fragile?
15:23
vagrantc: anything that requires running more than one command over the tunnel
15:23
<vagrantc>
warren: we already run several commands over the tunnel
15:23
<warren>
yes, but this is all for the purpose of keeping the current bad design?
15:24
I'm especially against a way to pass it in ldminfod
15:24
<vagrantc>
i don't see how the design is so infinitely bad.
15:24
warren: then why did you propose it as an option?
15:24
warren: kind of lead me on a wild goosechase with that one :P
15:25
<warren>
I only mentioned it, but I said I preferred the other one.
15:25
OK, the worst possible design is the current one, that looks on the client filesystem for the location.
15:25
<vagrantc>
yes, we couldn't do much worse than that.
15:25
<warren>
Slightly better is your proposed querying the server for a series of fallbacks.
15:26
It seems like we're willing to accept a great deal more complication in order to avoid something very simple.
15:26
going back to Oliver's reply
15:26Gadi has left #ltsp
15:26
<vagrantc>
and in your opinion, hard-coding a script path on the server seems like a good option. i don't think you'll get ogra or me to agree with that.
15:26
<warren>
"i dont like #2, it would mean you cant just connect to any ssh server at
15:26
all. ubuntu and debian offer CK functionallity through sshd (sshd uses
15:26
CK for evenry X connection made simly to make things like ssh -X
15:26
user@server time-admin works correctly) ... "
15:27
The first sentence is mostly inaccurate, while the second part is "We don't need it because we're shipping an ugly patch that upstream wont accept and is violation of the GPL thus I don't see a need to do this."
15:28
<johnny>
the script path can be changed at will tho
15:28
without updating the client image
15:28
<vagrantc>
johnny: no, warren's proposal is to hard-code it.
15:28
<warren>
vagrantc: because it is FAR simpler and requires a lot less code.
15:28
The only drawback is that you need to install a tiny shell script or even just a symlink from that location to your actual Xsession.
15:29
Since we agreed that we need a ldm-server split-out package anyway this isn't so bad.
15:29
<vagrantc>
warren: the accuracy of the first statement, you can't just connect to any ssh server, is absolutely true. it requires a script server-side.
15:29
<warren>
vagrantc: and that is FAR SIMPLER than the proposed client implementation
15:29
<vagrantc>
i agree that ldm-server is a good idea, but i disagree that it should be required.
15:30
<johnny>
warren, did i misunderstand? i thought it was that distros can decide what is supposed to happen without updating the chroot ?
15:30
and thus could be changed by editing that file
15:30
<vagrantc>
yes, but the client-side path to that file is hard-coded.
15:31
<warren>
vagrantc: that seems like a far saner way to do this than querying the server.
15:31
<johnny>
just as much as /etc/lts.conf is :)
15:31
or any other required file
15:31
<warren>
vagrantc: here's another issue
15:31
<johnny>
if we had autotools.. it would obviously handle PREFIX
15:31
<vagrantc>
warren: you can tell me it's saner and simpler all day long, but until you introduce new information, i still disagree that it's the only way to do it.
15:32
<warren>
vagrantc: say I do it the way I want, and you do it your way. Our client chroots are now incompatible in connecting to each other's servers.
15:32
johnny: I'm not following what you're talking about anymore
15:33
<vagrantc>
warren: only if you don't support it from ldminfod (which i guess you don't like, so maybe you wouldn't).
15:33
<johnny>
after what?
15:33
<vagrantc>
johnny: the difference between a hard-coded path to lts.conf is that it's a single environment, the thin-client. we're talking about cross-environment ... server-side and client-side.
15:34
johnny: a compatible interaction.
15:34
<warren>
vagrantc: do you know how to implement fallback detection of Xsession scripts?
15:34
<vagrantc>
warren: yes ... a simple for loop.
15:34
<johnny>
my fallback isn't even in there :(
15:34
<warren>
I'm especially uncomfortable with the fallback part
15:34
<johnny>
that reminds me..
15:35
<warren>
It is a losing game to hard-code every possible distro location in ldm
15:35
<vagrantc>
absolutely.
15:35
<warren>
vagrantc: perhaps a reasonable compromise is 1) prefer whatever ldminfod tells it 2) use whatever default you built into ldm client
15:35
no querying
15:36
<vagrantc>
warren: keeps it simpler.
15:36
<warren>
and this requires minimal code changes
15:36
<vagrantc>
warren: i would be ok with that.
15:36* vagrantc isn't sure how ogra would feel about it
15:36
<warren>
querying the server for fallbacks seems like extra lose.
15:36* vagrantc still wants LDM_SESSION configurable through lts.conf
15:37
<ogra>
dont we have LDM_COMMAND for that already ?
15:37
<warren>
ldminfod telling it sucks to me and I might not use it, but I'm not against my ldm following it if the server tells it to.
15:37
<ogra>
i would be ok with one hardcoded default and overriding ldminfod (thats actually what i thought)
15:38
<vagrantc>
ogra: hard-coded, or coded at build-time ?
15:38
<ogra>
build-time is fine
15:38
<warren>
build-time
15:38
<johnny>
ogra!
15:39
<warren>
to be clear, I mean
15:39
<johnny>
look at what wrath you have unleashed ..
15:39topslakr has quit IRC
15:39
<warren>
1) If ldminfod told it a Xsession location, use that. 2) Use whatever was hard coded into the client at build-time.
15:39
<vagrantc>
0) use LDM_SESSION if defined
15:39
<ogra>
would be /etc/X11/Xsession for me anyway ... i prefer to use the alternatives system server sided anyway ... and i suspect admins offering session choice are rather not the mass
15:39
<warren>
vagrantc: that already exists?
15:40
<vagrantc>
warren: yes.
15:40
i'll have to double-check the variable.
15:40
<warren>
vagrantc: wait, why do we need ldminfod to have this if LDM_SESSION already does?
15:40
<vagrantc>
but the concept exists
15:40
<ogra>
there was LDM_COMMAND as well with the pythin version
15:40
<vagrantc>
warren: because it's client-side
15:40
<ogra>
not sure thats still there
15:40
<warren>
vagrantc: it is exactly the same effect
15:41
<vagrantc>
warren: it's passed via lts.conf or something else.
15:41
<warren>
vagrantc: (both are misguided ways of telling the client what the client thinks it should do)
15:41
isn't lts.conf passesd by the server as well?
15:41
<ogra>
no
15:41
<vagrantc>
no
15:41
<warren>
huh?
15:41
you don't download lts.conf during client boot?
15:41
<ogra>
its only parsed from getltscfg
15:42
<warren>
(I thought you did so I added that to my client chroot.)
15:42
<vagrantc>
well, i guess you could say it's passed via the server, then.
15:42
<ogra>
right
15:42
it comes either from chroot or via tftp
15:42* vagrantc still prefers NFS ...
15:42* vagrantc confirms it's LDM_SESSION
15:42
<warren>
uh...
15:43
if LDM_SESSION already exists then I really don't see a point to adding it to ldminfod as well.
15:43
both lts.conf and ldminfod are coming from the server
15:43
you really don't need two different ways of doing this
15:43
<vagrantc>
lts.conf is coming from the root server, ldminfod could be on your login server.
15:43
<ogra>
LDM_SESSION is filled by ldm with the user choice from teh greeter gui iirc
15:43
<warren>
(and I think we should do neither, but I'm willing to accept keeping the existing one)
15:44
<vagrantc>
warren: then go ahead and put all locale and session information in lts.conf :P
15:44
<warren>
that isn't necessary
15:44
<ogra>
warren, lts.conf can come form a totally different server than your LDM_SERVER
15:44
<vagrantc>
i think there's use cases for all 3.
15:44
<ogra>
its also used for the multi server feature
15:44
<warren>
Th distinction between the 3 are confusing
15:45
<ogra>
(if you have a list of hosts to log in to)
15:45
<vagrantc>
warren: lts.conf comes from the root server. ldminfod should be running on all login servers, which could potentially be different distros.
15:45
<warren>
vagrantc: so, ldm could have multiple ldminfod files in the multi login server case, does it keep track of a different Xsession per server?
15:45
<ogra>
yes
15:45
(accoding to fgiraldeau and scottie)
15:45
<vagrantc>
well, currently it doesn't, but it would have to.
15:46
<warren>
ogra: vagrantc: and you don't think this couldn't be a LOT simpler with just a standard Xsession location on any valid login server? It is really not so onerous and eliminates a HUGE amount of extra code.
15:46
<vagrantc>
because right now we've just a hard-coded list of if-else
15:46
<ogra>
it queries ldinfod dynamically if your host selection changes
15:46bricode has joined #ltsp
15:47
<ogra>
warren, great, then lets default to /etc/X11/Xsession as we did the last two years :)
15:47
distros can just add a link :)
15:47
<vagrantc>
pfft.
15:48
<warren>
ogra: I'm thinking how technically sound that could be
15:48
<vagrantc>
warren: no less sound than what you're proposing.
15:48* ogra hopes warren got that he was joking
15:48
<warren>
That actually isn't a bad idea.
15:48
it is FAR simpler than all we're talking about
15:48
<vagrantc>
the only difference is some distro's would just work, and others would actually require a link.
15:48
<warren>
3 ways of configuring the same thing is a huge amount of lose.
15:48TheJoel has quit IRC
15:49
<warren>
No, there is a drawback to going back to /etc/X11/Xsession
15:49
<vagrantc>
a link, or a custom script
15:49
<warren>
while it would work for you today, you would lose flexibility
15:49
I would be able to put my own shell script there and do arbitrary things in the future possibly DIFFERENT from the standard Xsession script.
15:49
<vagrantc>
in case X12 ever happens?
15:49
<warren>
but you wouldn't be able to
15:50
<johnny>
why don't we jsut link in the server?
15:50
<ogra>
i wouldnt want to do things aside from the distro standard ....
15:50
<johnny>
/etc/ltsp/Xsession :)
15:50
to /etc/X11/whatever
15:50
<ogra>
... i uld change the standard :)
15:50
<vagrantc>
the debian Xsession scripts are quite flexible.
15:50
<ogra>
johnny, thats unneeded code duplication
15:50
<johnny>
i may not want my ltsp server to behave the same or use the same scripts
15:51
<ogra>
i *want* the debian/ubuntu Xsession script
15:51
<vagrantc>
ok, i've got to make dinner for some guests.
15:51
<johnny>
1 link is uneeded ?
15:51topslakr has joined #ltsp
15:51
<johnny>
ln -s /etc/X11/Xsession to /etc/ltsp/Xsession ?
15:51
<warren>
johnny: Fedora 9's /path/to/Xsession would be a wrapper that uses ck-launch-session then runs our Xsession, not just a link
15:51
<ogra>
so how do i log in to that redhat 4.2 server with the proprietary app that only runs on that old machine ?
15:52
<johnny>
so.. fedora doesnt' do that alredy?
15:52
<vagrantc>
ogra: write /etc/ltsp/Xsession ?
15:52
<ogra>
(the server where i have an ssh account but no write access)
15:52
... because its in the other dept.
15:53
<vagrantc>
well, then you're out of luck, because it doesn't use any of the known Xsession locations.
15:53
<ogra>
i want ldm to not lose that functionallity
15:53
<vagrantc>
it never had that functionality.
15:53
<ogra>
iÄm fine with adding all kind of nifty features ...
15:54
<johnny>
ogra, personally i think it will work out.. but right now i have to patch ldm to add my location :(
15:54
<ogra>
vagrantc, i can loh in to any ubuntu machine with sshd i have a pubkey and an IP for
15:54
<johnny>
since it's not in that list
15:54
<ogra>
s/loh/log/
15:54
<warren>
lts.conf LDM_SESSION exists now?
15:54
<johnny>
yes
15:54
i'm using it right now
15:54
<vagrantc>
ogra: yes, but not a redhat 4.2 system, which has a different path to Xsession.
15:54
<johnny>
since my location is not in that elseif
15:54
that's the only way i have it working
15:54
<ogra>
vagrantc, right, there i would have to use LDM_COMMAND/SESSION
15:55
<warren>
LDM_SESSION seems to be the redundant one then. ldminfod is the correct way to say "I'm not what you have defined in your build-time Xsession."
15:55
vagrantc: what the hell is redhat 4.2?
15:55
=)
15:55
<vagrantc>
but then it would set it globally for that thin-client.
15:55
<ogra>
why dont we just all agree that it needs improvement and be kind and give johnny his elseif for now :)
15:55
<warren>
vagrantc: if we need to go, let's talk about this later.
15:55
<johnny>
thanks ogra..
15:55
that's the rational thinking i was hoping for :)
15:55
<ogra>
warren, i got the CD upstrais :) and a book it was shipped in :)
15:55
<warren>
ogra: actually I'm not against that, I we seriously need to consider this more
15:55
<vagrantc>
warren: redhat 4.2 is actually the first distro i ever used seriously.
15:56
<ogra>
totally agreed
15:56
<warren>
vagrantc: red hat prior to 6.2 or so had no ssh
15:56
<ogra>
and in two weeks i will actually have time to really think about it :)
15:56
<johnny>
i ran debian in 1997
15:56* ogra too
15:56
<johnny>
until i couldn't get my video to work
15:56
<warren>
there was a time when openssh didn't exist, there was proprietary ssh only
15:56
<ogra>
but 4.2 is from '96 :)
15:56
<johnny>
then i went to win2k after awhile i got my first computer
15:57
<vagrantc>
warren: hmmm... i recall using ssh all over the place ... pre-6.x
15:57
<johnny>
i got debian via "boot magazine"
15:57
<warren>
vagrantc: are you sure it was openssh?
15:57
<vagrantc>
warren: no
15:57
<johnny>
go boot magazine
15:57
!
15:57
<vagrantc>
warren: but i didn't say it was either
15:58
<warren>
Anyone against just adding a fallback for johnny just for now? We're in disagreement of how to proceed and I really don't want us to hastily decide on a bad plan.
15:58
<vagrantc>
ok, give johnny the elsif, but it's got to be below /etc/X11/Xsession
15:58
<ogra>
warren, ++
15:58
<vagrantc>
we've already been bumped down two filesystem calls.
15:58
<ogra>
bah
15:59
<johnny>
that's fine .. i don't even have /etc/X11/Xsession :)
15:59
<warren>
OK, and we'll add Gentoo last
15:59
"save the best for last"
15:59
yeah
15:59
=)
15:59
<johnny>
beep
15:59
ok.. patch coming up in a few
15:59
<warren>
johnny: don't bother, just tell me the location, I'll add it now.
16:00
<vagrantc>
be sure to grab a recent ldm-trunk, as i just committed a translation a few minutes ago
16:00
<warren>
vagrantc: k
16:00* ogra takes a break ... to cool down his boiling brain
16:00
<johnny>
well.. now that you're going to.. i have to test one more idea of dberkholz's to make sure i'm picking the right one
16:00
the one i have now works, but it doesn't do what people expect it hink
16:00
so.. it'll be 20 min
16:00
<vagrantc>
didn't opensuse also have another location?
16:00
<warren>
vagrantc: I think we already have them
16:01mhterres has quit IRC
16:01
<vagrantc>
looking at the code, it's just got two.
16:01
<warren>
johnny: wait, this is for gentoo or what?
16:03Bengoa has quit IRC
16:04
<vagrantc>
yup, nobody every applied the patch for opensuse
16:04
so i guess debian/ubuntu is only down 1 filesystem call :)
16:05
but may as well apply that one too.
16:05
<warren>
Red Hat 7 was apparently the first time we shipped any ssh
16:05
<johnny>
warren, yes
16:05
<warren>
johnny: what is gentoo's location?
16:06
<johnny>
i have one, but dberkholz suggested a better location, i'm trying to test it now, but i need to fix up my chroot
16:06
so.. as soon as i do that, i'll let you know , i don't want to make you do it twice
16:06
<warren>
vagrantc: we could fork a background process/thread that finds the location of stuff while other things start, so you don't wait on blocking I/O? =)
16:06
<dberkholz>
johnny: oh, trying the Xsession in /usr/lib now?
16:06
<johnny>
yes
16:06
<dberkholz>
that one actually behaves normally like fedora's
16:06
<warren>
you guys have multiple!/
16:06
<johnny>
dberkholz, yeah.. what is the situation with that?
16:06
<dberkholz>
eh, i need to work something out.
16:07
i hate doing xdm stuff
16:08
<vagrantc>
how hard would it be to just add a ./configure flag ?
16:09
<warren>
./configure --xsession-location=/path/to/Xsession?
16:09abadger1999 has joined #ltsp
16:09
<vagrantc>
yeah
16:09
if it's not that difficult, why not just do it now?
16:09
<warren>
Do you agree to rip out all hardcoded locations then?
16:09
<johnny>
my autofoo is weak
16:10
<vagrantc>
warren: well, that's basically what i'm getting at.
16:10
<johnny>
so it would be harder for me :)
16:10
if you do that.. go ahead :)
16:10
<warren>
although
16:10
it would be rather broken for ./configure to fail without a parameter
16:10
so maybe we should have auto-detection
16:10
OTOH buildroot != server
16:11
<vagrantc>
right.
16:11
<johnny>
how can it detect what's on the server?
16:11
<warren>
but then we have the option coming from ldminfod
16:11
<johnny>
i don't think configure should autodetect based on a network service tho..
16:11
<warren>
johnny: only with clumsy and high latency "oops, that didn't work, let's try another"
16:11
<johnny>
that'd be kinda weird
16:12
<vagrantc>
warren: how about without parameter, ./configure defaults to using /etc/ltsp/Xsession
16:12slipttees has joined #ltsp
16:12
<johnny>
aha.. sanity returns :)
16:12
<warren>
ogra might have actually won me over with his bogus "proprietary app on redhat 4.2" argument
16:12
<vagrantc>
it's a distro-agnostic default, but doesn't require a given distro to use it.
16:12
<johnny>
lol...
16:12
<warren>
even though red hat 4.2 didn't have ssh
16:13
<johnny>
ogra is crazy yo..
16:13
nc ftw
16:13
<vagrantc>
but we don't support that anyways, because the current autodetection happens on the client.
16:13
<ogra>
yeah, i'm known to make up weird cornercase examples :)
16:13
<vagrantc>
that's a feature request.
16:13
<warren>
throw in a few buzzwords and convince venture capitalists
16:13joel_ has joined #ltsp
16:13
<ogra>
hehe
16:13
<johnny>
it's been ogracised
16:13
<warren>
haha
16:14
OK, I still think we should just add the client-side filesystem fallbacks for now
16:14
after Debian's location of course
16:14
We need to fully think this through
16:14
<johnny>
brb
16:14
<vagrantc>
i guess if it's not easy to add a configure option, then i guess that's the only thing to do.
16:14
<warren>
OTOH, it sounds like dberkholz and johnny aren't even sure which they want
16:15
vagrantc: none of us are good at autofoo
16:15
<vagrantc>
but it seems like if we're basically hard-coding the values, we need to figure out a way to hard-code them once.
16:15
<joel_>
Can I bother one of you for some answers on the ltsp system?
16:15
<vagrantc>
AHA!
16:15
set distro default in ltsp_config ?
16:15
<dberkholz>
i don't completely suck at autofoo
16:15
<warren>
vagrantc: so you can change it during runtime
16:16
err
16:16
without a rebuild I mean
16:16
<vagrantc>
nah, that doesn't work...
16:16
just set LDM_SESSION in lts.conf for everyone and be done with it.
16:16
:)
16:16
i've got to make dinner for some guests.
16:16
<warren>
I'm thinking LDM_SESSION should be continued to be supported (because it is bad to remove options)
16:16
but
16:16
we should deprecate it
16:16
because ldminfod is the correct way
16:17
especially with multiple login servers
16:17
<vagrantc>
no deprecation of a useful pfeature.
16:17
<warren>
It might not be useful when ldminfod can do it
16:17
you really don't want to confuse users by having both widely promoted
16:17
<vagrantc>
you might want a single terminal to use a different session than the others.
16:17
<warren>
argh
16:18
ok go, we'll talk later.
16:18
<ogra>
thats what lts.conf is for :)
16:18
<vagrantc>
exactly!
16:18
so don't deprecate it.
16:18
it's one line of code in ldm.c ... it's not hard to keep.
16:18
<warren>
ldminfod is only saying "if you login to me then you better use foople or it will not work"
16:19
<vagrantc>
no, it's saying "if you haven't been told to use, this is what you should use".
16:19
<ogra>
ldminfod is optional atm
16:19
<joel_>
On the LTSP system, I finally got a PXE boot to happen, but it only goes to a "busybox" prompt.
16:19slipttees has quit IRC
16:19
<warren>
lts.conf is for telling one client to do something possibly different from another client
16:19
<ogra>
it wont be used if you dont select a session
16:19slipttees has joined #ltsp
16:19
<vagrantc>
the session seleciton should still use LDM_SESSION ... at least on debian and ubuntu.
16:20
<warren>
it sounds like we need to step back and think about exactly what are all the ldm requirements
16:20
<vagrantc>
agreed.
16:20
<ogra>
right
16:20
we started that n seville
16:20
<warren>
too many different things we're talking about here
16:20
<ogra>
but during the cnversation more and more wierd requirements came up
16:20
<vagrantc>
so i guess the first thing is to get it working for opensuse and gentoo in the sub-optimal way.
16:20
<ogra>
francis ah lots and lots of extra reqs for millexterm
16:21
s/ah/has/
16:21
<dberkholz>
yay suboptimal!
16:21
<warren>
We need a picture of all requirements, then DELINEATE what everyone wants and do it in phases. If you aim too broadly it wont get done.
16:21
<vagrantc>
dberkholz: unless you can use your autofoo powers for the cause.
16:22slipttees is now known as slipttees_Disapp
16:22
<warren>
vagrantc: we need to know exactly what we want first
16:22
<vagrantc>
warren: true enough.
16:22
<ogra>
warren, well, all that was only with focus on debian and ubuntu back then ... and we didnt get it right yet
16:22
<warren>
I'm going to make a series of posts each with its own focused topic
16:22slipttees_Disapp is now known as slip_Disappointe
16:22
<warren>
so we don't get bogged down with many topics in the same thread
16:22
<ogra>
warren, expect the unexpected from francis :)
16:22
he has really weird usecases ...
16:23
but hue deployments ...
16:23
*huge
16:23
<warren>
Francis liked my proposal of an upstream Xsession location
16:23
<ogra>
well, he's canadian ...
16:23* ogra hides
16:25
<ogra>
hmm, canada jokes are somehow not the same wihout scottie around ...
16:25
<warren>
ogra: why don't people make fun of Americans?
16:26
ogra: we're so foolish now
16:26
<vagrantc>
warren: maybe they just don't do it in your company ...
16:26
<warren>
haha
16:26
<ogra>
warren, well ... thats rather bitter than funny
16:27slip_Disappointe is now known as DeviL_H4ck3r
16:27* vagrantc just spent two days in intense meetings in toronto
16:28
<ogra>
you travel a lot recently
16:29
<vagrantc>
yes, march and april has been an insanely vagrant couple months.
16:29
i intend to stick in one place from may till the end of july.
16:51deavid has quit IRC
16:54
<bricode>
Some Canadians leave the country until it's less cold...
16:54
For those of you that use Geode systems, I just finished testing a patch from Jordan Crouse...it allows libDDC calls without using the BIOS.
16:56
LTSP testing to commence in a few minutes.
16:59Guaraldo has left #ltsp
17:10mistik1 has quit IRC
17:12DeviL_H4ck3r is now known as slipttees
17:25shogunx has quit IRC
17:47artista_frustrad has joined #ltsp
17:51mistik1 has joined #ltsp
17:59mistik1 has quit IRC
18:24artista_frustrad has left #ltsp
18:31mistik1 has joined #ltsp
18:34abadger1999 has quit IRC
18:35abadger1999 has joined #ltsp
18:43shogunx has joined #ltsp
18:44beakburke has joined #ltsp
18:46
<beakburke>
anyone in here testing LTSP on fedora 9?
18:49joebaker has left #ltsp
18:50
<beakburke>
ls
18:56staffencasa has quit IRC
19:16
<dberkholz>
beakburke: pretty sure warren is =)
19:17
<beakburke>
he awol from chat right now??
19:18
<dberkholz>
00:18 [freenode] -!- idle : 0 days 1 hours 36 mins 59 secs
19:19
<beakburke>
most likely then. Is warren warrent T?
19:19
Looks like he is
19:21TelnetManta has joined #ltsp
19:30cpunches has quit IRC
19:34bricode has quit IRC
19:37GodFather has joined #ltsp
19:37beakburke has quit IRC
19:49bricode has joined #ltsp
20:05
<warren>
vagrantc: ogra: does your ltsp-setup start jetpipe on all clients even if you don't want it?
20:05
or need it
20:05
vagrantc: ogra: what exactly does jetpipe do anyway?
20:05
I mean concretely
21:04loather-work has joined #ltsp
21:07ryudo has joined #ltsp
21:13bricode_ has joined #ltsp
21:13
<vagrantc>
warren: i haven't used jetpipe, but basically it emulates a hp jetdirect port... kind of a semi-raw protocol for printers...
21:14
warren: takes data in on port 9100 and spits it out to the paralell and/or usb port
21:14
warren: other implementations are p910nd and lp_server.
21:15bricode has quit IRC
21:48
<ryudo>
vagrantc
21:48slipttees has quit IRC
21:48
<ryudo>
good night :)
21:49
do know the ltsp (hardy heron) have suport for un-partioned memory stickys ?
21:55
vagrantc ???
22:18spectra has quit IRC
22:31
<vagrantc>
ryudo: probably not
22:32
<ryudo>
in 4.2 thix fix exist but, why not in ltsp 5 ? :(
22:35
<vagrantc>
because we are not perfect.
22:36
make a good bug report, ideally with how to fix it, and it's much more likely to work.
22:37vagrantc has quit IRC
22:47
<warren>
sounds like jetpipe belongs in its own package.
23:26mccann has quit IRC
23:51ryudo has quit IRC