[17:35] mbiebl: of course, I keep forgetting when talking to Debian, that they're not actually programmers [17:39] w.r.t to startpar, you say? [17:40] maybe I'm just missing the context :-) [18:15] Keybuk: this weekend once I finish the tests for the state transfer patch, I'm going to try making 0.6 queue start/stop calls along with events [18:21] sadmac2: shiny [18:22] Keybuk: I try [18:23] Md: what is حَوّامتي مُمْتِلئة بِأَنْقَلَيْسون [20:05] wow [20:05] "make all" works in trunk for the first time in ages ;) [20:07] whee [20:11] you think that's rare? maybe you need to set up a buildbot [20:13] PovAddict: I'm not sure how that would help, other than flagging that it doesn't actually build [20:43] \o/ [20:43] | [20:43] / \ [20:43] nice [20:44] mbiebl: why are we celebrating? [20:44] [21:05:27] "make all" works in trunk for the first time in ages ;) [20:45] mbiebl: oh [20:46] Keybuk: does it only compile or actually boot ;-) [20:47] mbiebl: I haven't tested _that_ :) [20:47] I was being distracted by vim syntax highlight [20:47] oh [20:48] which doesn't work for me *cry* [20:48] are you using /etc/init/*.conf ? [20:48] I changed that bit [20:49] if I have a file loaded, how do I force vim to use upstart syntax for it [20:49] Keybuk: :source /path/to/upstart.vim methinks [20:50] set filetype=upstart [20:50] Keybuk: you're an emacs'er typically, aren't you? [20:50] yeah [20:50] Keybuk: ^^mbiebl has it right [20:50] mbiebl: I did that, and it turned off all the syntax highlighting ;) [20:51] Where did you install the upstart.syntax file into? [20:51] oh, I see [20:51] I'm an idiot [20:51] you called the file syntax_upstart.vim ;) [20:51] it needs to be upstart.vim in the syntax sub-directory [20:51] oh, yes [20:52] neat [20:52] I thought I had added that to the bug report [20:52] I probably didn't read it properly ;) [20:54] It needs a lot of polishing, but I wanted to wait with that, until the syntax format for 0.6 is finished. [20:54] It's actually a nice way to document all the keywords etc ;-) [20:57] time to run make: real 0m45.070s [20:58] time to run make check: real 10m35.532s [20:58] on my crappy laptop, i.e. [21:02] that's not bad [21:02] it passed then? :) [21:02] It did, yes [21:04] * sadmac2 wonders if make check is -j safe [21:04] sadmac2: don't think so [21:05] Keybuk: I'd imagine the dbus stuff would get hairy [21:05] though automake may not even allow that [21:05] it might build them in parallel but run them in series [21:05] indeed, that's exactly what it does ;) [21:05] so it's "safe" in the sense that it's ignored [21:55] Keybuk: I think in the future we shouldn't have dates in the code names. It confuses me. [21:55] I keep thinking its some other sort of information [21:56] heh [21:56] but it was the release date! :) [21:56] Keybuk: why were you John Masters? [21:57] (for who the Sega Master System is named btw. Fun fact) [21:58] do you not listen to the lkml summary podcast? [21:59] Keybuk: no. I subscribe to the list... I'm a little behind :P [21:59] pick one at random and listen for a few seconds [21:59] then you'll understand [22:00] cool [22:00] and now I go home [22:33] well, holy crap, it booted [22:51] Keybuk: can confirm [22:51] running it too ;-) [22:51] trunk? [22:52] yep [22:52] are you using the shiny new default jobs? :) [22:52] I do [22:52] runlevel handling is a bit broken [22:53] oh, what did you find? [22:53] when booted, runlevel should be "N 2" [22:53] not "S 2" [22:53] I think it should be S 2, no? [22:53] you've come from single-user mode into runlevel 2 [22:53] no, don't think so [22:54] I know 0.3 said N 2 [22:54] but I always thought that was a bug ;) [22:54] personally I wish Ubuntu packages used upstart events instead of init.d scripts :) [22:54] it's not like you stopped after rcS and then switched runlevel manually to 2 [22:55] hmm maybe [22:55] otherwise, you have the problem that when you enter runlevel 2 [22:55] just change the env RUNLEVEL= in rc-sysinit.conf [22:55] all stop links are executed [22:55] should be just [22:55] env RUNLEVEL= [22:55] ohhh [22:55] maybe it wasn't a bug then [22:56] e.g. I have apache and mysql installed, but disabled [22:56] and when entering runlevel two, those two are stopped first [22:56] before the actual boot continues [22:57] ok [22:57] I changed that [22:57] now when I boot into single-user mode first, it's [22:57] N S [22:57] (it was S S before, which I agree is a weeeee bit odd) [22:57] ke230434 * sadmac2 wonders if make check is -j safe [22:57] N S for single user mode is correct I think [22:57] .NOTPARALLEL: or something equivalent should be used if it’s not. [22:58] But I think, the rcS stage during boot shouldn't be considered "single user mode" [22:58] (Now that i read further, “something equivalent” is done. :-P) [22:58] yeah, I mean with "single" [22:58] then when I go from single to multi-user (^D) I get S 2 [22:58] that looks about right too [22:59] let's try booting straight into multi-user [23:00] now it's N 2 [23:00] and the system boot record was still written \o/ [23:00] cool [23:01] what I also find a bit weird, is that the jobs are actually named job.conf [23:01] sure [23:01] e.g. in status etc [23:01] hm [23:01] they are? [23:01] are you sure you installed the 0.6 util directory? [23:02] rc.conf stop/waiting [23:02] actually, that shouldn't make a difference [23:02] I see [23:02] etc [23:02] rc stop/waiting [23:02] which revno are you at? (upstart & libnih) [23:02] 1160 / 966 here [23:02] 1155/963 [23:03] I guess it's time for an update then ;-) [23:03] oh, yeah [23:03] you're missing the revs that dealt with the .conf issue [23:04] 1161 has the rc-sysinit fix [23:04] compiling ... [23:21] keybuk: How are you planning to handle mounting with 0.6 in karmic? [23:21] I wish i had managed to write the mount daemon, but i haven’t. [23:22] by using a mount daemon :) [23:22] :-) [23:23] Any code yet? [23:28] no, that's my next thing to do really [23:30] Keybuk: noticed one more minor issue [23:30] when I use startpar, switching to runlevel 6/0 will also stop the service [23:30] which were already disabled in 2 [23:31] for debugging, I added a pre-start script echo $(runlevel) end script to rc.conf [23:31] then this problem magically goes away [23:34] odd, why does it do that? [23:35] isn't echo $(runlevel) a little superfluous? :p [23:35] runlevel would be enough there I actually have a simple runlevel call ;-) [23:36] just wanted to make it clear, that it's debugging output [23:36] /etc/init.d/rc uses the $runlevel and $prevlevel env variables [23:36] and passes that to startpar [23:37] it sets those from $RUNLEVEL and $PREVLEVEL though right? [23:37] checking... [23:38] runlevel=$RUNLEVEL [23:38] that should be set [23:38] previous=$PREVLEVEL [23:38] [ "$previous" = "" ] && previous=N [23:39] weird [23:39] set -x that script? :p [23:40] weird is, that simply calling runlevel magically "fixes" it [23:41] that is weird [23:41] runlevel doesn't do anything ;) [23:49] I added a runlevel to /etc/init.d/rc [23:49] (when runlevel and previous are set) [23:49] and on reboot it shows N 6 with startpar [23:50] but of course it should show 2 6 [23:51] what does it show without startpar? [23:51] fwiw I also get an assertion error from this runlevel call, when it's called during rcS [23:51] without startpar it shows 2 6 [23:51] weird thing is, startpar is called much later in /etc/init.d/rc [23:51] do you have a core for that? [23:52] or at least the assertion message? [23:52] Keybuk: I'll need to reboot for that and, wait a sec [23:55] runlevel:error.c:319 Assertion failed in nih_error_get: context_stack != NULL [23:55] Aborted [23:59] got a core for that?