[00:00] Keybuk: any other thoughts? [00:01] not offhand [00:02] Keybuk: should I keep plowing on this or go back to fiddling with 0.10 features? [00:03] which 0.10 features were you thinking of? [00:04] Keybuk: I started from the top. was ripping out start on and stop on in favor of just plain ol' on. [00:04] which got me thinking about this because changing the config file format was miserable. [00:04] on is something quite specific though [00:05] its the biggest leap toward 0.10 though because its changes are most destructive. There's more bits coming off. [00:05] I'd say that the most destructive will be reorganising the jobs/instances stuff [00:08] that's not really destructive as much as additive [00:08] if you remove start on/stop on you've removed the mechanism that automatically starts and stops jobs. [00:08] Then you just have to replace it. [00:08] I don't agree [00:09] I intend to change the very meaning of start and stop [00:09] and since the ideas haven't been written down yet [00:09] it's far too early to jump in and start recoding things [00:10] what are these changes? [00:11] the idea that instances are created automatically [00:11] based on the while condition and matching instances there [00:11] start/stop simply change an existing instance - never create them [00:12] that's still not much change. All that means is "kill job_class_start" [00:12] I think it's a fundamental change [00:12] I know you're eager to get coding on things [00:12] but right now, I can't honestly think of any particular bits that need changing ye [00:13] * Keybuk prefers to write things down before starting [00:15] You English and your native tongue. I have no use for anything written down in such an overcomplicated language. C is a much better way of expressing ideas :) [00:16] no, blasting in and rewriting things is a very good way to get yourself in a mess [00:18] the mess is leftovers from the early stages. that comes from not blasting hard enough :) [00:19] ? [00:21] nothing wrong with blasting and rewriting if you commit. [00:21] don't get attached to the old code. and don't wait. don't let a function sit 5 minutes past its interface mildly annoying you. [00:32] Keybuk: do you leave 1 or 2 lines between functions? It seems to vary throughout the code [00:32] depends on the relationship between the functions [00:32] 1 line if the functions are realted [00:32] 2 if not [00:32] ah [10:51] Silly question but are the *buntus going to start adding more standard upstart events ? [10:52] After Upstart 0.10 [10:52] for instance it would be nice if there were some like "suspend-started" or "resume-finished" etc [10:52] hmm when is that ? [10:53] As soon as it’s ready. ;-) [10:53] will that be for koala or whatever it is called ? [10:53] fair enough [10:53] Dunno [10:57] Whats happening for 0.10 that standard events need to wait for it ? [11:02] It will implement enough functionality for the entire system to boot with native jobs. As soon as that point is reached, there’s incentive to think of “standard events”. [11:03] (and standard states) [11:06] Sounds good. thanks for the info. I'll just keep implementing my own events until then :) [18:50] * Keybuk has working daemon supervision now [18:51] Keybuk: what'd you do? [18:52] magic stuff [18:52] Keybuk: its bad for your sceptum to do that stuff [18:55] tracing a trace program is hard. [18:55] keesj: the Linux Journal had a piece about you and the fork/ptrace stuff in the "wat's up on kernel front" [18:55] keesj: stop talking to yourself. [18:56] :p [18:56] oops [18:57] Keybuk: what kind of magic? [18:59] O hai, ion_ [19:00] sadmac: greetings from your work self, home self! [19:01] * sadmac2 heads to class [19:07] sadmac2: dark magic [19:08] keesj: got a link? [19:11] searching.. [19:22] failed: scanning... [19:29] Keybuk: http://box.mmapps.net/lm_100_upstart/ where 003 is the part I was talking about