[10:47] <alan_g> alf__: anpok I trust there's no objection to TAing https://code.launchpad.net/~alan-griffiths/mir/use-compare_exchange_weak-correctly/+merge/236557?
[10:48] <alf__> alan_g: let me take a quick look
[10:53] <alan_g> alf__: sure, np
[11:01] <alf__> alan_g: ok, TAed
[11:02] <alan_g> alf__: thanks
[11:22] <anpok> alan_g: no objections oh that was too late
[11:22] <alan_g> anpok: 8)
[12:13] <kdub_> it was a long while ago that this written, but I'm wondering why the mir.protobuf.Buffer.fds_on_side_channel (and similar) is not in the wire protocol
[12:14] <kdub_> seems like the wire protocol should be the raw bytes and fds, and then it gets further parsed into the mir_protobuf.proto concepts
[12:48] <alan_g> kdub_: It is probably that the fd passing depends on the message type - and that's just raw data in the wire protocol.
[12:49] <alan_g> And then the fd passing got added to one message after another.
[12:51] <alan_g> What you suggesting could well make sense. (Any problems will only become apparent from trying to do it.)
[12:51] <kdub_> we can determine that an invocation needs to get fds from the side channel based on the message type, i'm grappling with how we can tell how many fds we need to collect before dispatching the method
[12:51] <kdub_> I have a smallish MP i'll propose and see what people think about it I suppose
[12:52] <kdub_> (its been a while since I considered the wire protocol), but its also strange that our events get jammed in the rpc response message type too
[12:53] <alan_g> that comes from thoughts about batching messages that were never fully realized
[12:53] <kdub_> ah, okay. that satisfies my curiosity about it
[12:54] <alan_g> It will all go away when RAOF re-implements it using Wayland
[12:55] <kdub_> :) sounds good to me
[15:54] <robotfuel> kgunn: ping, I noticed this https://bugs.launchpad.net/ubuntu/+source/unity-system-compositor/+bug/1376324 crash on errors.u.c (has not happened during long running tests) can you triage it for me?
[15:59] <kgunn> camako: can you pick a voluteer for that one ^
[15:59] <kgunn> looks like it'd be childs play
[15:59] <camako> kgunn, ack