[12:35] <vasilis> hey,just one small question,how can i change the runlevel in upstart?
[12:52] <mbiebl> Keybuk: there are problems, compiling upstart on hppa
[12:52] <mbiebl> Seems, hppa does not support __thread
[12:53] <mbiebl> http://experimental.ftbfs.de/fetch.php?&pkg=upstart&ver=0.2.7-1&arch=hppa&stamp=1163110346&file=log&as=raw
[12:58] <mbiebl> Maybe we should abandon the usage of __thread in libnih
[01:07] <_ion> ...or use them if the compiler and system supports it?
[01:07] <_ion> +only
[01:11] <mbiebl> or so
[01:25] <Keybuk> mbiebl: hppa supports threads in feisty :p
[01:28] <mbiebl> Keybuk: Does it need gcc >= 4.2 on hppa?
[01:30] <Keybuk> not sure whether it's gcc or glibc
[01:34] <mbiebl> I'll ask the porters, how to deal with that appropriately. Probably it's just a simple #ifdef
[01:40] <Keybuk> probably
[01:41] <Keybuk> just define __thread to nothing
[01:42] <theCore> Keybuk, have you just presented upstart to Google folks?
[02:22] <Keybuk> theCore: no, why?
[02:23] <theCore> Keybuk, oh, ok. because I listened to Mark's talk and heard you talking about upstart
[02:23] <theCore> or maybe it wasn't you ...
[02:25] <Keybuk> it wasn't me
[02:25] <Keybuk> not during Mark's talk, anyway
[02:26] <theCore> it hard to tell who is speaking with VoIP :/
[02:26] <theCore> ok, thanks
[02:29] <Keybuk> I was talking in a little room just now
[02:31] <theCore> Upstart is really becoming one the big features of Ubuntu
[02:33] <thom> muh? it has the potential to do so; it doesn't actually do anything bar busting  cryptdisks yet ;-)
[02:38] <theCore> thom, :)
[02:38] <Keybuk> The Core was a cool movie
[02:39] <Keybuk> all the annoying characters die
[02:40] <mbiebl> One of the worst movies I've ever seen ;-)
[02:40] <Keybuk> oh, why?
[02:42] <mbiebl> Absurd story, bad effects, poor playing
[02:42] <Keybuk> but it's such fun
[02:43] <mbiebl> Well, if you see it from that angle, sure ;-)
[02:43] <Keybuk> how can you not love a film that actually called its magical element necessary for the plot "unobtanium" ?
[02:44] <Keybuk> heh @ imdb
[02:44] <Keybuk> # Incorrectly regarded as goofs: Since almost all of the "science" in the movie is entirely erroneous, we are prepared to accept that the movie's universe *must* have entirely different rules - it's the only possible explanation. It's just for fun.
[02:45] <mbiebl> hehe, had to laught about the physics too
[02:45] <theCore> Keybuk, are you making fun of my nick ? :P
[02:46] <Keybuk> theCore: no
[02:46] <Keybuk> I do actually like that movie
[02:46] <theCore> I do too
[02:48] <theCore> Keybuk, I just wondering, was it related to my nick at all?
[02:49] <Keybuk> well, theCore; The Core
[02:49] <theCore> oh
[02:50] <Keybuk> my brain is low on caffeine
[02:51] <theCore> once you choose a nickname, you're really stuck with it
[02:52] <theCore> I would had chosen something more sane, if I had expected that passion for free software
[02:53] <theCore> anyway...
[02:54] <theCore> (FYI, the Core is from the game Total Annihilation, which I used to love before coming to the world of Linux)
[02:55] <Keybuk> ahh
[02:56] <theCore> ever heard of that game?
[02:58] <theCore> you're making me remember a whole bunch of good souvenirs
[02:58] <thom> TA++
[02:59] <Keybuk> nope
[02:59] <Keybuk> my gaming knowledge probably slightly predates yours
[02:59] <Keybuk> (as my usual hostnames will attest)
[03:00] <theCore> I know there is an open source implementation out there
[03:00] <thom> total annihilation is probably entirely responsible for me failing my second year of uni
[03:01] <theCore> thom, lol
[03:01] <thom> many games went longer than 24 hours
[03:02] <theCore> thom, did you ever gotten into unit creation?
[03:02] <thom> nah
[03:03] <theCore> me neither
[03:03] <theCore> I remember that you had to script the units in COBOL
[03:04] <theCore> ugh!
[03:28] <theCore> thom, btw, did you ever heard about TA Spring (http://taspring.clan-sy.com)?
[05:37] !christel:*! This is your captain spealing, we hit a bit of turbulence there. We're now cruising at approximately 13,000 ft. Thank you for flying freenode air and have a good day ;)
[09:25] <AlexExtreme> hmm
[09:25] <AlexExtreme> how would i handle critical errors in upstart jobs? i.e, if udev fails, the boot process should be stopped
[09:25] <AlexExtreme> how would i do that?
[09:28] <phsdv> all jobs waiting for udev/started will never start, so booting stops automatically?
[09:28] <AlexExtreme> that's a point
[09:29] <phsdv> maybe you want to start sulogin?
[09:29] <AlexExtreme> that would be good
[09:30] <AlexExtreme> maybe i could raise a "boot-failed" event and have sulogin start on boot-failed ?
[09:32] <phsdv> yep, maybe that would help
[09:32] <AlexExtreme> k
[09:36] <phsdv> what about udev/failed event when a script exits with an error code. Does that exists already?
[09:36] <AlexExtreme> i don't know
[09:43] <AlexExtreme> Keybuk: when is the replacement-initscripts spec planned to be implemented? i mean, will it be done for feisty
[09:50] <AlexExtreme> Keybuk: the reason I ask is that I think it would be better to wait for ubuntu to switch to upstart jobs, then i have some idea of the best way to do stuff. i want to know how long i'd need to wait, if you see what i mean
[10:52] <mbiebl> Keybuk: hi
[11:04] <Keybuk> AlexExtreme: yes, feisty
[11:06] <mbiebl> Keybuk: what version of binutils do you ship in feisty?
[11:07] <mbiebl> The problem with TLS on hppa seems to be a binutils issue which is said to be fixed in 2.16.8
[11:07] <mbiebl> Argh, I mean 2.18
[11:08] <Keybuk> mbiebl: it was fixed within the week
[11:08] <mbiebl> You mean binutils?
[11:09] <Keybuk>  *** 2.17.50.0.6-0ubuntu1 0
[11:09] <Keybuk>         500 http://archive.ubuntu.com feisty/main Packages
[11:09] <Keybuk>         100 /var/lib/dpkg/status
[11:10] <mbiebl> It's likely that TLS support will be a requirement for etch+1. As far as I could find out, hppa and m68k are currently not yet ready in Debian.
[11:11] <mbiebl> I was wondering, why __thread is needed at all, as upstart itself is not multithreaded.
[11:11] <Keybuk> mbiebl: libnih is used by more than upstart
[11:11] <mbiebl> Would you say it is safe to just remove it for upstart?
[11:11] <Keybuk> can you not just patch it out?
[11:12] <Keybuk> easy to make an autoconf macro for that
[11:12] <mbiebl> I'm not such an autoconf guru.
[11:13] <mbiebl> Would it be easy to write a test for it and check it during configure?
[11:13] <Keybuk> yes
[11:13] <mbiebl> Would I need a test compile for that?
[11:16] <Keybuk> yeah
[11:16] <Keybuk> look at AC_C_CONST for example
[11:16] <Keybuk> it #defines const to nothing if it does't exist
[11:19] <mbiebl> Were is that macro defined?
[11:19] <mbiebl> s/Were/Where/
[11:19] <theCore> Keybuk, does libnih is used elsewhere than in upstart?
[11:23] <mbiebl> http://www.nabble.com/libgomp-crash-fix-t2595290.html
[11:23] <mbiebl> Looks good, doesn't it?
[11:24] <mbiebl> Keybuk: could you include that check directly upstream?
[11:24] <Keybuk> mbiebl: yeah
[11:52] <mbiebl> Keybuk: I'll prepare patches against upstart and libnih. Shall I open a LP bug report or send it to you privately?
[12:05] <Keybuk> LP is fine
[12:05] <Keybuk> link it to the debian one