/srv/irclogs.ubuntu.com/2014/06/03/#ubuntu-mir.txt

kdub_fixing 1299977 is oddly cathartic00:00
kdub_also, how much code should we share between our examples and our production code? I say ideally 'nothing' and the examples should only be based on the public headers00:01
=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
=== fginther` is now known as fginther
=== chihchun_afk is now known as chihchun
=== tedg is now known as ted
racarr__Morning!13:51
alan_gracarr__: Afternoon here13:52
alan_gracarr__: got time to make a couple of trivial fixes to https://code.launchpad.net/~mir-team/mir/cursor-spike-lightning-round-fix-input-area-contains/+merge/220111 - so we can land it?13:53
alan_ganpok_: racarr__ here's a real easy one to review: https://code.launchpad.net/~alan-griffiths/mir/fix-1325602/+merge/22184613:54
racarr__alan_g: Thanks! as soon as I finish breakfast I will13:56
alan_gdednick: What do you think of:14:14
alan_gs/trusted_participant_starting/participant_added/g14:14
alan_gs/trusted_participant_stopping/participant_removed/g14:14
dednickalan_g: i have no problem with that.14:15
* alan_g makes it so14:15
=== marrusl_ is now known as marrusl
=== dandrader is now known as dandrader|afk
=== dandrader|afk is now known as dandrader
kdub_we're in this funny position where we have an api/abi between the production code and examples/15:42
kdub_my particular problem is that both the demo compositor and the default compositor need a renderer and some occlusion detection15:43
alan_gWhat's funny about production code having an api/abi and example code using it?15:44
kdub_well, i'd be okay if the api/abi was only the stuff in include/ of course15:45
kdub_but in our examples, we're recommending modifying the production code to suit the purposes of the examples15:45
alan_gThat's odd15:45
kdub_particularly mc::filter_occlusions_from and mc::GLRenderer15:46
* alan_g suspects the answer lies in bzr blame15:47
kdub_well, the answer lies in how we're sharing GLRenderer15:47
kdub_like, its useful because its the default, but do we want to share the code that is useful to the mir defaults in the hope that shell writers can make use of it too?15:48
kdub_I say we don't, if you override a default, you have to write a full alternative15:48
alan_gkdub_: I'm not familiar enough with the current state to know how reasonable it is. (Can one just wrap the default and change things that way?)15:50
kdub_well, one could wrap the default, but we allow overridding various internally-called functions in GLRenderer15:51
alan_gkdub_: I mean through the public api15:52
kdub_right, like take the_gl_renderer() and wrap the default GLRenderer with shadow/frame painting15:54
alan_gMy feeling is that we don't have enough use-cases (yet) to drive a correct design, and that we've got [premature] generalization in the wrong places.15:56
kdub_right, and I think the right place to move to overridding is the DisplayBufferCompositor (which just has a bool composite() function)15:57
kdub_instead of Renderer which is funny interface with begin(), render(), end(), set_rotation(), etc15:57
kdub_I just want to figure out if I should try to share mc::filter_occlusions_from with the demo shell, or just write a demo shell occlusion algorithm that doesn't use production code15:58
alan_gI'm not voting on the "correct" answer. I just know the code is hard to adapt every time we try supporting something new. It needs good test cases and a refactor.16:00
racarr__Im voting for decorations and shadows in the scene :p16:02
alan_gkdub_: I don't think it hurts to share that algorithm.16:03
kdub_it doesn't hurt to share, but hurts a bit to expand it to think about fancier decorations :)16:03
alan_gkdub_: you were happy with the previous version (but CI wasn't). Could you check: https://code.launchpad.net/~alan-griffiths/mir/rework-test_client_library.cpp-mk2/+merge/221854 - thanks16:09
kdub_okay16:09
racarr__alan_g: Oh updated lightning-round btw16:20
alan_gracarr__: top approved an hour ago btw16:25
racarr__alan_g: Aha, thanks :)16:35
alan_ganpok_: OK to land https://code.launchpad.net/~alan-griffiths/mir/fix-1325602/+merge/221846?16:54
=== chihchun is now known as chihchun_afk
=== alan_g is now known as alan_g|EOD
=== dandrader is now known as dandrader|lunch
anpok_alan_g|EOD: yes17:41
racarr__</lunch>18:20
racarr__mwahaha I replaced my keybaord18:28
racarr__with my laptop18:28
racarr__and set up synergy for mouse sharing18:28
racarr__its quite nice18:28
racarr__ (+ keybooooooard)18:28
racarr__...though apparently with some key repeat glitches18:28
racarr__digging in to the emulator bug now...its a little weird... (https://bugs.launchpad.net/ubuntu/+source/android/+bug/1319582)18:30
ubot5Ubuntu bug 1319582 in Mir "emulator: 'Failed to start RenderThread' after opening/closing applications" [High,Triaged]18:31
racarr__I dont know where the fix is but it cant be in mir/qtubuntu, etc, because how are they supposed to clean up18:31
racarr__a clients eglWindowSurface (which in the emulator model has driver/host ssssssssssssssside reeeeeesources)18:32
racarr__...wow this keyrepeat ttttthing is great18:32
racarr__I mean its like you allocate an eglContext and then crash and nothing leaks, presumably the kernel side component of the driver cleans up any hardware resources18:34
racarr__so this is an emulator driver bug that android has worked around somehow (or we are misusing the emulator)18:34
racarr__Who ever thought "apt-get source android" would be a thing18:37
racarr__failed to fetch...hash sum mismatch?!?!18:43
=== dandrader|lunch is now known as dandrader
=== ted is now known as tedg
racarr__typedef SomeSortOfSmartPtr<type> TypePtr is an ugly pattern :p19:57
racarr__I think I tracked down the emulator bug...and have a strong guess thats what lead to it.19:57
AlbertAanaybody else with canonical irc connection issues?20:40
racarr__nah. launchpad connection issues though20:41
=== chihchun_afk is now known as chihchun
=== Sarvatt_ is now known as Sarvatt
=== ValicekB_ is now known as ValicekB
kgunnhey there...for anyone wanting to lend a hand in testing23:59
kgunnwe're trying to promote our mir 0.2.023:59

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!