[10:43] <hugo303> Fiveby5184
[10:45] <hugo303> *slaps his head, hard, and changes password
[10:46] <Keybuk> d'oh
[12:30] <rikard> Does anyone know how to get upstart 0.3.9 to work with multiple dependencies?
[12:39] <Keybuk> it doesn't
[14:26] <sadmac> Keybuk: I didn't know there was going to be a 0.3.10...
[14:26]  * sadmac would have written a lot more patches
[14:27] <Keybuk> heh
[14:27] <Keybuk> I didn't necessarily plan one
[14:27] <Keybuk> but assert() bugs are always worth fixing
[14:27] <sadmac> damn, you should have held it up. I could have fixed 10 minor annoyances over the weekend.
[14:28] <sadmac> Red Hat BZ is full of 3-liner "flag foo is missing from shutdown" type patches
[14:28] <sadmac> well, its full of the complaints anyway
[14:30] <Keybuk> those don't seem worth fixing ;)
[14:31] <Keybuk> once you cave in and agree an option should exist, you have to keep it forever
[14:38] <sadmac> Keybuk: you did say "SysV compatible" when you released. RH Engineering is holding you to that quite rigorously.
[14:38] <sadmac> Keybuk: also, you could have cleaned up/taken the "preserve state across re-exec" patch
[14:39] <sadmac> fedora carries it
[14:39] <Keybuk> no I didn't
[14:39] <Keybuk> that's obvious by the fact Upstart doesn't read /etc/inittab <g>
[14:40] <Keybuk> preserve state across re-exec -> then that would have to be supported by 0.5 and 0.10
[14:40] <sadmac> Keybuk: I seriously hope we're going to support that...
[14:40] <Keybuk> support which?
[14:43] <sadmac> preserving state across reload
[14:43] <Keybuk> yes
[14:43] <Keybuk> and once I do, that API will never be broken
[14:43] <Keybuk> which is why I'm being slightly hesistant of writing down that API
[14:47]  * Keybuk tries to figure out his branch mess
[14:47] <Keybuk> *ahem*
[14:47] <sadmac> Keybuk: I think introducing it into 0.3.* wouldn't be harmful. 0.5+ breaks from that in so many ways anyway
[14:48] <sadmac> by 0.10 state transfer between minors wouldn't even make sense. 0.10 state doesn't even look that much like 0.3.9 state
[14:57] <Keybuk> exactly my point?
[14:58] <sadmac> Keybuk: so why not carry the 0.3 patch and retire the release in style?
[14:59] <Keybuk> better question, why carry the patch?
[15:00] <Keybuk> carrying it will suddenly bless that API, and there will be bugs when it doesn't work in future
[15:00] <Keybuk> it doesn't cost you anything to apply it in Fedora after all
[15:00] <sadmac> you can close with ENOTSUPP
[15:00] <Keybuk> but it must be supported
[15:00] <Keybuk> the patch is in the release
[15:00] <Keybuk> see what I mean
[15:00] <Keybuk> if I apply the patch and release with it, I'm declaring it supported
[15:00] <sadmac> which it will be, in the 0.3.9 series only
[15:01] <sadmac> across minors, you can just say "no"
[15:07] <Keybuk> hmm, it's not an unconvincing argument ;)
[15:07] <Keybuk> if you could file a bug (target 0.3), clean up and attach the patch (making sure it has test cases) - I'll do a 0.3.11 with it
[15:09] <sadmac> sweet. I might compile a list of other things that might suit a 0.3.11. You can line-item those when I get them (they'll all be comparatively tiny)
[15:10]  * sadmac wonders what else we're carrying in CVS...
[15:10] <sadmac> shouldn't be much...
[15:11] <sadmac> oh wow. 9 patches.
[15:12] <sadmac> force-on-shutdown-reboot is a genuine bugfix. you'll want that.
[15:12] <Keybuk> sure, file bugs on them
[15:12] <sadmac> will do.
[15:12] <Keybuk> use the "Target to release" thingy to set them as 0.3
[15:12] <Keybuk> (you can also tag 0.5 and trunk that way too)
[15:13] <sadmac> Keybuk: is ubuntu built on gcc43+ yet?
[15:13] <sadmac> we're still carrying a patch to add a header file that needs
[15:13] <Keybuk> yes
[15:13] <Keybuk> limits.h?
[15:13] <Keybuk> that one went into 0.3.10
[15:13] <sadmac> yeah
[15:13] <sadmac> cool
[15:13] <Keybuk> as did an amd64 fix
[15:14] <sadmac> Keybuk: oh, want to roll our events.5 manpage into a general release?
[15:14] <Keybuk> sure
[15:14] <sadmac> cool
[15:15] <ion_> CVS⁈
[15:15] <sadmac> I'll have all this filed by saturday (earlier if my date stands me up tonight)
[15:15] <sadmac> ion_: don't remind me
[15:19] <sadmac> ion_: Fedora packages are in CVS. We have one branch per release. Inside each branch is an rpm specfile and all the patches we apply (upstream tarballs are kept separate in a lookaside cache and pulled in when we build)
[15:20] <sadmac> its actually fairly easy to work with. we have a makefile that does the hard stuff.
[15:21] <sadmac> cahnge specfile, copy in patch, commit, make tag, make build
[15:22]  * Keybuk tries to make util build again
[15:23] <Keybuk> shutdown.c: In function ‘shutdown_now’:
[15:23] <Keybuk> shutdown.c:490: error: too few arguments to function ‘nih_dbus_proxy_new’
[15:23] <sadmac> phantastiq
[15:28] <Keybuk> builds now ;)
[15:29] <sadmac> 5 minutes. Not bad. But can you do it in space?
[15:30] <Keybuk> I _AM_ in space
[15:31] <sadmac> The boss took you along on his vacation this year?
[15:31] <Keybuk> I wish
[15:32] <ion_> I am in time.
[15:32] <Keybuk> the most we get for good behaviour is a ride in the jet
[15:32] <sadmac> I think Shuttleworth jokes are the equivalent of saying "your mom" to an Ubuntu dev.
[15:32] <sadmac> at least the spaceboy ones.
[15:33] <Keybuk> Your CEO is so dull he worked for Delta 
[15:33] <Keybuk> etc.
[15:35] <sadmac> Jim's actually a lot more fun than I thought. Not as intimidating as Szulik, but he drinks the Kool-aid, which is nice
[15:36] <sadmac> runs rockbox on his iPod
[15:36] <sadmac> look forward to the release of Red Hat Enterprise Music Bus soon
[15:39] <sadmac> echo -e '#include<pthread.h>\n#include<unistd.h>\nmain(){pthread_t
[15:39] <sadmac> t;pthread_create(&t,NULL,pause,NULL);pause();}'|gcc -o t -x c - -pthread
[15:39] <sadmac> thats... a new way to specify a reproducer in BZ...
[15:44] <Keybuk> BZ?
[15:45] <Keybuk> oh, BugZilla
[15:46] <Keybuk> damnit, the 0.3 code is easier to read than the 0.5
[15:50] <sadmac> Keybuk: what's the cause?
[16:03] <Keybuk> sadmac: of the code being easier to read?
[16:04] <sadmac> Keybuk: rather of it being harder. Is there a particular disease?
[16:04] <Keybuk> I don't like the way blocked stuff is done in 0.5 basically
[16:05] <Keybuk> then again, at least it work s:p
[16:05] <Keybuk> ...with respawn of running while post-start process
[16:05] <Keybuk> test:job_process.c:1090: Assertion failed in job_process_terminated: job->state == JOB_POST_START
[16:05] <Keybuk> \o/
[16:05] <sadmac>  |
[16:05] <sadmac> \
[16:06] <sadmac> there's a regular on fedora-devel-list who's name is listed as King InuYasha
[16:06] <sadmac> I can't tell if he's serious or not
[18:45] <Keybuk> now, if distcheck would just pass ...
[18:47] <sadmac> Keybuk: what's up with it?
[18:48] <Keybuk> just taking its time
[18:48] <Keybuk> there's no reason it shouldn't pass
[18:48] <Keybuk> but I'm paranoid
[18:48] <Keybuk> and out-of-tree builds hate me
[18:49] <ion_> A process too slow? Throw more hardware at it. :-)
[18:50] <Keybuk> /home/scott/co/upstart-0.5/upstart-0.5.2/_build/util/../../util/initctl.c:349: undefined reference to `job_get_name_sync'
[18:50] <Keybuk> see
[18:50] <Keybuk> HATRE
[18:50] <Keybuk> HATE
[18:50] <sadmac> carpe odium
[18:53]  * Keybuk waits again
[18:53] <Keybuk> Too Many Test Cases
[18:54] <sadmac> Keybuk: we could probably parallellize a lot of them if we thought about how
[19:12] <ion_> Next up: 0.10. :-P
[19:13] <sadmac> Keybuk: we went to GPLv3? I thought your license preferences tended the other way
[19:15] <sadmac> Keybuk: you rewrote nih_dbus_tool in expat?
[19:15] <sadmac> *or rather in C
[19:23] <keesj> http://paste-it.net/raw/public/j16d3e1/ same kind of error  undefined reference to `dbus_message_iter_abandon_container'
[19:23] <keesj> that one looks more like a dbus thing actualy
[19:45] <mbiebl> Is that related to the D-Bus patch that Scott referenced in his announcement email?
[19:46] <mbiebl> Keybuk: what bad things will happen, if this patch is not applied?
[19:47] <Keybuk> sadmac: my licence preferences don't apply to Upstart
[19:47] <Keybuk> that's up to Canonical ;)
[19:47] <Keybuk> keesj: see NEWS
[19:47] <Keybuk> mbiebl: Upstart won't combile, obv :)
[19:48] <sadmac> Keybuk: oic
[19:48] <mbiebl> Keybuk: ah,ok. So keesj' problems are indeed related to this missing dbus patch
[19:49] <Keybuk> you can replace _abandon_ with _close_
[19:49] <Keybuk> but then libdbus will randomly assert() on you
[19:51] <Keybuk> you can remove the lines, but then you'll get a memory leak
[19:51]  * sadmac is near to cleaning up the hideous locking in linux's wait*() syscalls
[19:51] <sadmac> no more quantum-critical sections!
[19:58] <Keybuk> cisco in "whinging about licensing" again shocker
[19:59] <keesj> I have the same problem . I feel very sorry
[20:02]  * keesj goes and reads the GPLV3 again
[20:11] <Keybuk> keesj: why?
[20:12] <Keybuk> What is it about v2->v3 that you don't like?
[20:14] <keesj> I don't think is about liking or not. at least at work we gplv3 will be very hard(read impossible) to sell.
[20:14] <Keybuk> you don't ship any current versions of FSF software?
[20:14] <Keybuk> what version of gcc do you use?
[20:15] <keesj> our company has booth hardware and things like maps they really like to sell
[20:15] <Keybuk> you can sell hardware with GPLv3 software on it
[20:15] <keesj> we use recent gcc like 4.2 but that doesn't pose restrictions on the software we create 
[20:16] <keesj> I can sell a close source project compiled with gcc so that is not a good example
[20:16] <Keybuk> gcc 4.2 is under the GPLv3
[20:16] <Keybuk> which version of coreutils do you use?
[20:17] <keesj> don't know something relased with the code sourcery we use let met check
[20:17] <Keybuk> CodeSourcery the company/products?
[20:17] <keesj> the toolchain
[20:17] <Keybuk> they're partners of Canonical, and certainly include GPLv3 pieces
[20:18] <keesj> nope , well perhaps lgpl
[20:19] <Keybuk> (you know that GPLv2 software can't be linked with LGPLv3 right? :p
[20:19] <Keybuk>  so every single piece of GPLv2+ software on a system linked to an LGPLv3 library is automatically GPLv3 :p)
[20:20] <keesj> well I guess I need to find an other job :p
[20:20] <Keybuk> seriously though
[20:21] <Keybuk> why the concern?
[20:21] <Keybuk> we do work in the embedded and mobile spaces
[20:21] <Keybuk> no GPLv3 trouble there
[20:25] <keesj> dear Mister GSM provider please provice us the source code for your GSM stack so we can abuse your network, and make your device not FCC compliant
[20:26] <sadmac> oh boy do people ever have some funny ideas about what this license means
[20:26] <Keybuk> you don't need the source any more than you do in GPLv2
[20:26] <Keybuk> you can ship GPLv3 code in the same system as commercial code
[20:26] <Keybuk> just as you can with GPLv2
[20:27] <keesj> no it's not about source it's about freedom for the devloper !
[20:28] <Keybuk> huh?
[20:28] <sadmac> keesj: what freedom do you think has gone missing?
[20:29] <keesj> well the "tivoization" part for example or wanting keep control over what software runs on the embedded device
[20:30] <sadmac> keesj: what's wrong with that? I paid for the damn device.
[20:30] <sadmac> and LGPLv3 strikes all of that
[20:30] <sadmac> explicitly
[20:31] <sadmac> brb kernel
[20:38] <keesj> Well quite frankly I will check the license of the binutils and such but for the moment assume GPLV3 is not OK
[21:16] <sadmac> Keybuk: I don't see your patch in linus yet. Who said they took it?
[23:27] <joshk> is there a way to print something to tty0 even during a quiet boot?
[23:27] <joshk> i have this /etc/rc.local file that takes a while to run
[23:27] <joshk> i've tried: stderr, and log_*_msg from /lib/lsb/init-functions, and nothing shows up
[23:28] <joshk> i guess upstart is filtering out all of stderr and stdout. There's no way to work around that?