[11:43] <_ion> Note to self: when Keybuk returns, ask him whether it would be a reasonable idea to integrate something like http://irccrew.org/~cras/security/data-stack.html to libnih.
[12:42] <Keybuk> yay
[12:42] <Keybuk> we have little upstart icons on beta lp now
[12:51] <mbiebl> url?
[12:52] <Keybuk> mbiebl: you have to be a member of the launchpad beta-testers group to see it
[12:53] <mbiebl> ah, ok 
[12:53] <Keybuk> http://quest.netsplit.com/~scott/beta-lp-bugs.png
[12:53] <Keybuk> ("private screenshot") :p
[12:54] <mbiebl> just read the note ;-)
[02:37] <_ion> The launchpad beta is cool.
[02:37] <_ion> Btw,
[02:38] <_ion> ti124335 < _ion> Note to self: when Keybuk returns, ask him whether it would be a reasonable idea to integrate something like  http://irccrew.org/~cras/security/data-stack.html to libnih.
[02:40] <Keybuk> will look at that in a sec
[02:45] <_ion> last black
[03:09] <Keybuk> eh?
[03:10] <_ion> Forgot the / from '/last black'. :-)
[03:10] <Keybuk> heh
[04:00] <asdaf> does anybody knows a reason why upstart can't create @/com/ubuntu/upstart socket?
[04:01] <asdaf> *know
[04:02] <asdaf> release 0.3.5
[04:04] <Keybuk> almost certainly because it's already running?
[04:04] <Keybuk> what's the error?
[04:04] <Keybuk> it won't just say that it can't create it, it'll tell you why it couldn't create it
[04:07] <asdaf> i noticed it because initctl fails connecting to it
[04:08] <asdaf> it doesn't show in /proc/net/unix
[04:08] <Keybuk> then you're not running upstart
[04:08] <Keybuk> upstart would've failed to even boot
[04:08] <Keybuk> lsof /dev/initctl
[04:29] <asdaf> i don't have lsof (it's an embedded) but looking at /proc/1/fd dir there is no reference to a unix socket
[04:29] <asdaf> same system on another box works fine
[04:29] <asdaf> i'm wondering what can make the difference
[04:33] <Keybuk> can you give me the ls -l of /proc/1/fd
[04:39] <Keybuk> _ion: doesn't seem to give anything that nih_alloc doesn't already do?
[04:40] <Keybuk> frame = nih_alloc (NULL, 0);
[04:40] <Keybuk>   /* alloc everything with frame instead of NULL */
[04:40] <Keybuk> nih_free (frame)
[04:40] <Keybuk> unless he's combining it with a slab allocator, of course
[04:42] <asdaf> Keybuk, pvt?
[04:43] <_ion> keybuk: You'd need to pass the frame variable to function calls etc.
[04:44] <Keybuk> asdaf: sure
[04:44] <Keybuk> _ion: true
[04:45] <Keybuk> it'd make the whole "allocate temporary stuff that we throw away again" thing easier
[04:45] <Keybuk> that's a little cumbersome with nih_alloc
[04:45] <_ion> keybuk: I haven't actually read his implementation, but he says "No memory fragmentation".
[04:45] <Keybuk> *shrug* probably a slab allocator to return the things you ask for
[04:46] <Keybuk> (that's where you allocate memory in pages at a time, and split it into managed chunks)
[04:46] <_ion> Yeah
[04:46] <_ion> He also says "FAST, most of the time allocating memory mean only updating a couple of pointers and integers. Freeing the memory all at once also is a fast operation", which sounds like slab, too.
[05:24] <AlexExtreme> mmm
[05:24] <AlexExtreme> nice icons on launchpad beta :)
[05:44] <Keybuk> yeah
[05:46] <AlexExtreme> hmm, i think i made a mistake with my job files... i installed the ones that I wrote on my host system, it didn't quite work ;)
[06:00] <Keybuk> oops
[06:07] <_ion> Sigh
[08:03] <_ion> keybuk: Btw, what might the syntax of the scripting language built around syscalls you've envisioned look like?
[08:09] <_ion> Hahaha, the comment in the latest libnih commit :-D
[08:12] <Keybuk> probably a bit like Perl. now I think more <g>
[09:18] <Keybuk> need to work out what's in the box for 0.3.9 I guess
[09:29] <Keybuk> http://www-128.ibm.com/developerworks/library/l-boot-faster/index.html
[09:31] <AlexExtreme> nice#
[09:31] <AlexExtreme> *nice