[06:26] <duflu> Wow, platform-api needs some love. It has merge proposals untouched going back 18 months
[06:37] <anpok_> duflu: you misunderstood the mp.. though we already settled with the other.. the instantiation is pushed to the code using the configuration.. and in case of internal code will not be exposed as a symbol and in case of somebody outside of mirserver using it.. it will reside there
[06:37] <anpok_> so it is not part of the abi
[06:44] <duflu> anpok_: Isn't that still DefaultServerConfiguration? That's the leaf of the class hierarchy
[06:45] <duflu> Using templates still means just as many real functions get created. If they're that easily moved then we probably could have done it without templates
[06:45] <anpok_> hmm people could just interact with serverconfiguration and insert or query whatever they need
[06:45] <duflu> Maybe we have already -- I'm not going to start looking at the other branch
[06:46] <anpok_> and thats kind of the ugly side
[06:46] <anpok_> as it is harder to figure out what is provided and who pushes inside..
[06:47] <anpok_> hmm
[06:47] <anpok_> i added the templates to have it type safe
[06:47] <anpok_> otherwise i would have to expose the shared_ptr<void> based interface
[06:48] <duflu> anpok_: Doesn't matter now if something else has been chosen. And for my own sanity I need to not look at it any more till the sprint at least
[06:48] <anpok_> well I am not entirely convinced of both solutions..
[06:48] <duflu> Too many MPs
[06:49] <anpok_> but alans one is the better one for now
[06:49] <anpok_> untill we get wiser..
[06:49] <anpok_> and maybe still then
[06:52] <duflu> anpok_: Wiser still to stay out of it all ;)
[12:59] <anpok_> alf__: alan_g: WrappingBuilder instead of Wrapper?
[13:01] <alan_g> Why? It uses more words to convey the same concept.
[13:05] <anpok_> because it is only the function returning a wrapper.
[13:10] <alan_g> wrapper: 1. a person or thing that wraps.
[13:12] <alan_g> (You're thinking "2. something in which a thing is wrapped.")
[13:13] <alan_g> BuildingFunctor/WrappingFunctor :^)
[13:13] <anpok_> alan_g: you mean it wrapps the functor that builds the type
[13:14] <alan_g> no. There's a functor that builds and a functor that wraps
[13:17] <anpok_> ok
[13:17] <anpok_> brb
[13:33] <kdub_> AlbertA2, with thte frameskipping/krillin issue, that doesn't occur with overlays on correct?
[13:56] <alan_g> kdub_: does render_overlays do anything useful on desktop (I get an exception "trying to createGBM buffer with unsupported pixel format")
[13:57] <kdub_> alan_g, it could be used to try out bypass, but in the current form thats not easy to do
[13:57] <kdub_> and that issue would be something else to figure out what's going on
[13:58] <kdub_> it needs some sprucing up, mostly to render more interesting things, and perhaps have more indicative output of what's going on
[13:58] <kdub_> in terms of accepting/rejecting an optimization
[13:59] <alan_g> kdub_: OK (I wanted to check I hadn't broken - but if it seems already broken)
[13:59] <alan_g> thanks
[14:02] <kdub_> yep, no problem
[14:05] <kdub_> https://bugs.launchpad.net/mir/+bug/1381552
[14:05] <kdub_> seems it needed a wishlist bug
[14:41] <AlbertA2> kdub_: right I have not seen it happen with overlays
[14:42] <kdub_> AlbertA2, i've been watching for a while with overlays disabled, and I havent seen it
[14:42] <AlbertA2> kdub_: egltriangle?
[14:43] <kdub_> yes
[14:43] <AlbertA2> I wonder if we got a new vendor binary blob
[14:43] <kdub_> possible
[15:04] <robotfuel> greyback_: I have a better stacktrace in https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1377332
[15:05] <greyback_> robotfuel: thanks
[15:07] <greyback_> robotfuel: the spinner was animating?
[15:07] <robotfuel> greyback_: yes there is a link to video on my chinstrap account
[15:07] <greyback_> robotfuel: yep I saw that from the video
[15:07] <greyback_> just wnated to confirm
[15:08] <greyback_> robotfuel: is it still hung, or have you rebooted it?
[15:08] <robotfuel> greyback_: I've re-imaged with todays image :/
[15:08] <robotfuel> greyback_: I think I still have log files
[15:09] <greyback_> robotfuel: nah, I would have asked for you to install a few more debug packages, as there's some symbols missing which could help
[15:09] <robotfuel> greyback_: I can do it next time, which packages?
[15:09] <greyback_> I'm just about to set off some more random testing myself, so with luck I'll get it
[15:10] <greyback_> robotfuel: dandrader mentioned them in comment 22: https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1377332/comments/22
[15:11] <robotfuel> greyback_: ah I didn't install those, just the qt dbg packages, I'll make sure to install those on the next one.
[15:11] <greyback_> robotfuel: thanks. If you get another hang, please ping me
[15:12] <robotfuel> greyback_: actually the phone hasn't been imaged yet, I can still install those.
[15:13] <greyback_> robotfuel: well if you get hang, we cna attach gdb, try generating stacktrace, see what debug symbols we need, install them and try gdb again
[17:11] <kdub_> tricky to see the best place to put the fd size description
[17:11] <kdub_> in the ipc layer
[17:11] <kdub_> the header, the wire protocol, or the mir protocol...
[17:12] <kdub_> we currently have it in the mir protocol, which seems wrong
[18:35] <robotfuel> greyback_: are you still around? I have another ui freeze
[18:35] <greyback_> robotfuel: yep
[18:36] <greyback_> robotfuel: please attach gdb to the frozen unity8 process, and do a backtrace
[18:36] <robotfuel> greyback_: https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1377332/+attachment/4237481/+files/gdb.txt
[18:36] <robotfuel> greyback_: done :D
[18:38] <robotfuel> greyback_: this one is on mako instead of krillin like before
[18:39] <greyback_> robotfuel: ok. In gdb, would you mind entering "info threads" and pasting the output to me
[18:40] <robotfuel> greyback_: the backtrace unlocked the freeze
[18:40] <greyback_> robotfuel: that isn't possible, but more likely something timed out and unblocked the rendering