[09:08] mbiebl: which debian packages? [11:35] the speed difference between "initctl list" and "sudo initctl list" is kinda interesting [14:12] Keybuk: the packages you uploaded to karmic [14:20] mbiebl: of upstart? [14:20] yeah [14:20] I didn't look in the Debian package [14:20] was there something I missed? [14:22] Keybuk: for one, there was upstart-job already in their [14:22] and you merged the debs into one [14:22] (and some smaller stuff) [14:23] oh, I didn't know you'd done the upstart-job thing yet [14:23] I just stuck a quick-and-dirty shell script in there for now, until you did :) [14:25] maybe we should communicate better then ;-) [14:26] why the merge of the packages? [14:27] imho keeping e.g. the tty job files and the core upstart package separat made sense [14:44] the separation we had didn't make much sense [14:44] it was kinda arbitrary [14:47] ok, I'm confused ;) [14:48] you were saying I merged the Debian packages [14:48] did you mean that I merged the binary debs together [14:48] or merged with the Debian package (compared to the Ubuntu one) [14:48] I meant, merged the debs [14:48] ahh [14:48] right, yeah [14:48] we talked about doing that in Ubuntu a while back [14:49] since most things will ship their own upstart jobs, and we'll try and push them upstream [14:49] even util-linux will ship the jobs for the gettys [14:49] nothing would end up in those two packages [14:49] so it would be much easier to just drop system-services when util-linux does that [14:50] and with merging upstart-compat-sysv back into upstart (since shutdown, etc. are "native" now), they kinda all went away [14:50] really? doesn't make much difference surely? [14:50] util-linux will ship /etc/init/getty.conf [14:51] you don't have to follow that in Debian of course ;) [14:51] that's up to you [14:51] what did you mean about upstart-job btw? [14:54] Keybuk: my point is not so much about the merge itself but having known beforehand what you are planning [14:55] you were in the room at UDS when we talked about it ;) [14:55] sorry, then I completely missed that part [14:55] when was this discussed, the the binary packages will be merged? [14:55] in the packaging policy session I think [14:55] the room with the glass wall [14:56] the getty bit specifically was in the karmic kms console one though [14:56] I honestly can't remember that [14:56] rationale: [14:56] system-services, currently only contains getty [14:56] we want to merge those into one instance job [14:57] now we've discovered that we probably need to dynamically create ttys anyway [14:57] so it all makes sense to move to util-linux [14:57] (leaving that package empty) [14:57] startup-tasks is already empty [14:57] there is no point arguing about that ;-) [14:57] things like running udev belong in udev, setting hostname belongs in util-linux, etc. [14:58] upstart-compat-sysv, we discussed here ages ago that it didn't make sense to hang those tools out to dry [14:58] arguing about startup-tasks, I mean [14:58] they got moved from compat/sysv to util in the main source package (pre 0.5 iirc) [14:58] and now the default recommended jobs are shipped in the source package as well [14:58] especially when upstart gains native lsb jobs, it becomes increasingly blurry [14:59] having everything in one binary package makes the migration easier too [14:59] the missing piece is something to migrate custom changes from /etc/event.d to /etc/init [14:59] and clean up after [14:59] if all the resulting files are in one package, that's easier :p [15:01] Keybuk: slightly off topic: you said 0.10/1.0 will continue to support 0.6 job definitions, right? [15:02] right [15:03] ah. that'll make life easier. [15:03] much as I'd love to delete all that code :) [15:03] my preliminary packages on http://debs.michaelbiebl.de/upstart/ [15:03] the idea being that people should feel safe about updating to 0.6 now [15:03] and using it as a stable release [15:04] while 0.10 is a development release that they can experiment with without worrying [15:04] had the maintainer scripts updated to support the conffile migration [15:04] mbiebl: ah, I deliberately didn't do that because 0.3 and 0.6 job files aren't directly compatible [15:05] Keybuk: hehe, I only migrated the ttyS [15:05] you need to do more than just copy them [15:05] for the same reason I on rm the old if unmodified [15:05] what do you if modified? [15:05] keep a copy [15:05] in /etc/event.d ? [15:06] yeah [15:06] Keybuk: did we ever increase the strictness of the 0.6 job format? does having two start stanzas still just silently take the second one? [15:06] sadmac2: yes [15:07] Keybuk: yes to which question :P [15:07] sadmac2: what you said is true; the last of all dup stanzas is used [15:07] Keybuk: unfortunate, but not much to be done about it. [15:08] Keybuk: regarding the merge: I guess you should also add Conflicts [15:08] why unfortunate? [15:08] mbiebl: already did that ;) [15:08] oh, not in -1 [15:08] which I looked at [15:08] Keybuk: it should at least give a warning. That behavior is almost certainly not reflecting what the user intended when they wrote the job definition. [15:09] And why the Provides? [15:09] sadmac2: it followed what most other configs do [15:09] e.g. [15:09] SomeOption on [15:09] SomeOption off [15:09] SomeOption on [15:09] Keybuk: if most other configs jumped off a bridge... :P [15:11] sadmac2: I'd push you off ;) [15:12] Keybuk: touche [15:12] I'd give you a portal gun first [15:13] Keybuk: potential release name: "Thank you for helping us help you help us all." [15:13] * sadmac2 <3 GlaDoS [15:23] Keybuk: about upstart-job [15:24] imho it makes sense to map force-reload to restart [15:25] I also strip away any leading SK?? from the basename [15:26] so it can be started e.g. by /etc/rc2.d/S50cups [15:27] Keybuk: Idea is, that packages can potentially start now to ship upstart job files [15:27] without strictly having to got from bottom up [15:28] e.g. cups could ship a job file today [15:28] which then is not started by events [15:28] but by sysv [15:29] which imho is pretty neat, as we don't to strictly synchronize when to convert which packages [15:33] Keybuk: did you look at the state transfer patch? [17:16] Keybuk: I think you can drop the Provides from the Ubuntu package [17:16] and I also think it doesn't make sense to ship the nih-dbus-binding tool man page [17:17] mbiebl: we need those in Ubuntu the way that ubuntu-meta works [17:17] how's that? [17:18] ubuntu is installed by packages that depend on others [17:18] ubuntu-minimal depends on those packages provided [17:18] hm, doesn't the latest ubuntu-minimal not only depend on upstart? [17:19] latest yes, but not when upgrading from jaunty to karmic when released [17:19] the installed one will still depend on upstart-compat-sysv [17:19] providing it makes upgrade ordering easier :) [17:20] so update-manager will fall over if those provides are not there? [17:20] no [17:20] it'll just get pessimal ordering [17:22] what's pessimal? [17:24] ok, seems to be the opposite of optimal ;-) [18:36] keybuk: I’m thinking of fixing some typos etc. in the 0.6.0 manpages. How are the lines wrapped? I don’t seem to find a consistent maximum line width or equivalent.