[04:32] <aent> hey... I'm not too familiar with upstart currently, but I'm currently working on a project where I need something to manage running processes from the project, start them, do dependency management, basically what upstart does, but on a higher level
[04:33] <aent> is upstart's code suitable for that? or is it tightly integrated into being specifically for init? is there something better for what I'm trying to do?
[04:34] <aent> its basically needs to monitor processes to make sure they don't die, if they do, bringing down the dependencies and bring them back up on the proper order after the other processes have started, allow different processes to be started and stopped, etc
[05:02] <sadmac> aent: upstart doesn't do dependencies in the traditional sense. monit might be more what you need
[05:02] <aent> hmm ok
[16:40] <sadmac2> Keybuk: what is the purpose of the allocater being pluggable in nih_alloc?
[16:43] <Keybuk> future stuff
[16:44] <sadmac2> Keybuk: I was toying with rewriting nih_alloc, and that would have to go
[16:45] <Keybuk> rewriting it how?
[16:46] <sadmac2> Keybuk: replace malloc + header with mmap + free list + trailer
[16:46] <sadmac2> basically re-implementing malloc inside libnih (in such a way that we get information out of it that lets us cull most of the header overhead)
[16:47] <notting> is there a reason to not use the system malloc? desperately needed in signal handlers?
[16:48] <Keybuk> is the header overhead really that worrysome?
[16:48] <sadmac2> Keybuk: for smaller strings it can be
[16:49] <sadmac2> Keybuk: I think I can get it down to 1 or 2 bytes for most blocks
[16:50] <sadmac2> and I can do the multi-parent deal
[16:53] <Keybuk> this doesn't seem worth optimising at this point? :p
[16:53] <sadmac2> meh. prolly not
[20:50] <sadmac2> Keybuk: I think in libnih head you somehow reverted the dbus timeout change by mistake
[20:51] <sadmac2> that'd explain why reverting looked so weird
[20:52] <Keybuk> it's entirely possible
[20:52] <Keybuk> in fact, I think you're right
[20:52] <sadmac2> it'd also explain why the timeout issues seemed to be resurfacing
[21:07] <sadmac2> also build RPM packages.
[21:07] <sadmac2> wrong window/button/appendage
[21:07] <sadmac2> general desk disaster
[21:10] <sadmac2> Keybuk: are you thinking about rolling another libnih? I think what we have now in terms of foreach is far better than what's there
[21:10] <sadmac2> it does something a little unexpected, but nothing idiotic.
[21:10] <sadmac2> also I think a 0.5.1 upstart might be welcome, if only to update the manpages