[08:05] Hello happy people. [08:06] I just discovered that the upstart cookbook uses some horrible anti-patterns in its examples [08:06] for dir in `ls /var/lib/queues` [08:11] Anyways, the question I'm here for actually, other than, how do I fix the examples in that cookbook and send you patches, is: Can I change setuid/setgid *per* instance? [08:28] Use more quotes: http://apaste.info/12l8 [14:13] jodh, are you working on the file bridge? and if so, is there already some code? [14:13] I also wanted to know if there is html documentation for libnih. would be nice to have this at developer.ubuntu.com [14:14] toabctl: yes, I am. I'll push some code somewhere soon. I forget who (xnox?), but someone @ UDS offered to auto-gen some dev docs using doxygen as NIH uses the KDE commenting-format. [14:15] jodh: s/KDE/gtk-doc/ [14:15] xnox: close :) [14:16] toabctl: jodh: yeah I did generate gtk-doc locally for both, it doesn't look very nice (formatting errors), tried to clean up and integrated and it broke again. [14:17] xnox, would be nice to autogenerate the doc with the package. so build-dep would be gtk-doc [14:17] toabctl: I appreciate this isn't to everyones tastes, but I've configured vim for ctags for nih+dbus so you can just CTRL+] to jump to the actual NIH code (and doc of course). [14:18] toabctl: jodh: filed as bug 1087269 subscribe if you are interested =) [14:18] toabctl: yeah, i know. But gtk-doc needs correct bootstrapping with autoconf first which I didn't quite do right the first time around. When it works, I will propose a merge ;-) [14:19] xnox: subscribed. [14:20] xnox: Keybuk still owns NIH upstream, but if you can put an NIH branch with doc changes somewhere we can work on a new release branch and see if it gets his blessing :) [14:20] I'm still waiting for STL man pages :) [14:22] jodh, how do you want to implement user sessions? with a new upstart instance for every user? or only one upstart instance which handles all the stuff? [14:22] jodh: STL? stereolithography file format? [14:22] toabctl: it's all here: https://wiki.ubuntu.com/FoundationsTeam/Specs/RaringUpstartUserSessions [14:23] xnox: C++ standard template library :) [14:23] toabctl: but in short, 1 instance per user. Makes for a much cleaner design and avoids nasty DoS possibilities. [14:23] toabctl: make that 1 instance / user *session* :) [14:24] jodh, something different is mentioned here: https://blueprints.launchpad.net/ubuntu/+spec/desktop-q-upstart-session-requirements (see bottom of page) [14:24] jodh, sometimes imho it's very confusing with alle the blueprints, wikis and mailinglists. I never know what's the right source [14:25] toabctl: that's the quantal blueprint, not the raring one. [14:25] jodh, so half a year later all is different ? :-) [14:25] jodh, some should at least mark the blueprint as outdated/invalid [14:26] and point to the latest information for the topic [14:26] toabctl: it is marked "Implementation: Deferred". [14:27] toabctl: it's not that difficult - just go to https://blueprints.launchpad.net/ubuntu/$release (ie raring) and see what the community is working on. [14:28] jodh, ok. and the blueprints from blueprints.launchpad.net/upstart ? [14:32] toabctl: first time I've seen that page :) Those bp's refer to "upstream Upstart projects" whereas we are driving changes via the Ubuntu project (even though they actually land in upstream Upstart first). [14:38] :) [21:50] slangasek: when you have a minute, I'd like an opinion on https://code.launchpad.net/~stgraber/ubuntu/raring/libnih/libnih-dbus-as-type/+merge/138582 [21:51] slangasek: it appears to work here, diffing upstart's generated code shows that it's only fixing the one variable I added, but I won't pretend to understand all the potential impact of that change ;) [21:57] stgraber: hmm, I'm not familiar enough with this code myself to be a particularly good reviewer [21:58] I can dive into it, but it'll take me a while. Maybe it's better to have jodh review? [21:58] slangasek: ok. I wasn't sure how much of libnih you knew about. I can wait till tomorrow for jodh to review. [21:59] slangasek: at least I have a local build with that fix so I can now write tests for my dbus branch without having upstart die on me :) [22:06] slangasek: spoke too soon, looks like I'll need more magic in nih, the generated code segfaults in upstart though the nih testsuite itself passed. I guess it's missing a test with an empty array... === You're now known as ubuntulog [22:06] mmk :) [22:25] right, so I managed to fix the generated code so that the upstart testsuite works, now to fix the generator to give me something that roughly matches my manually patched file :)