=== chihchun is now known as chihchun_afk
=== chihchun_afk is now known as chihchun
alan_galf_: any opinion? https://code.launchpad.net/~alan-griffiths/mir/more-surface-resize/+merge/256912/comments/64140708:33
=== cimi is now known as Cimi
=== pete-woods1 is now known as pete-woods
=== mibofra is now known as Guest34273
=== willcooke_ is now known as willcooke
=== olli is now known as Olli
=== alan_g is now known as alan_g|reboot
alan_ganpok_: any opinion? https://code.launchpad.net/~alan-griffiths/mir/more-surface-resize/+merge/256912/comments/64140711:46
=== alan_g is now known as alan_g|lunch
=== dandrader is now known as dandrader|afk
=== alan_g|lunch is now known as alan_g
alan_gkdub AlbertA: any opinion? https://code.launchpad.net/~alan-griffiths/mir/more-surface-resize/+merge/256912/comments/64140713:06
* kdub looks13:07
=== dandrader|afk is now known as dandrader
=== chihchun is now known as chihchun_afk
alan_gcamako: I just came across some TODO comments in src/server/frontend/buffer_stream_tracker.h - are they a legacy of your (now abandoned) effort to change the buffer ownership logic?13:59
=== ara is now known as Guest54483
kdubalan_g, I think that comment was mine at some point14:14
kdubalan_g, and I think it was about how if the frontend code held a shared_ptr somewhere in there, we could re-bury the acquire/release (wrapped by swap_buffers() nowadays)14:15
kdubbut its kinda dusty, as we've gotten rid of TemporaryClientBuffer14:16
alan_gkdub: I was wondering if the comments should go14:16
anpok_kdub: do you want to have another look on the resubmitted MP here: https://code.launchpad.net/~andreas-pokorny/mir/dispatchable-event-hub/+merge/25644514:17
kdubalan_g, yeah, it probably can14:17
kdubanpok_, okay14:17
kdubanpok_, with that, what was the lockup that was happening in the test?14:19
alan_gkdub: for avoidance of doubt you're referring to both the first two TODOs in that file. (The third one looks wrong too, but is different)14:20
kdubalan_g, ah, yeah, i was referring to the third one14:21
kdubwhat? a mir file with more than one todo in it? :)14:21
kdubwell, the part about the shared ptr is part about it14:21
kdubdoh, mispoke, reorienting...14:22
kdubalan_g, so the 1st and 2nd todo are about how originally, i thought a unique_ptr should get deposited there, and released when the client message came back, so its pretty old, as the BufferQueue hasn't quite evolved that way14:23
anpok_lockup? the problem the first time since landed was that no initial scan happened and then that after the scan there would be blocking call to epoll_wait, instead of just testing for pending events14:25
* alan_g thinks he should leave it to kdub to revise these comments.14:25
anpok_*the first time this MP landed..14:25
kdubalan_g, mk, easy to do today14:26
anpok_hum I guess the new input dispatcher will need another round.. so I guess I have to fit the old one meanwhile14:26
* kdub is prodding around the {mf,mc}::BufferStream interface today anyways14:26
kdubanpok_, re lockup https://bugs.launchpad.net/mir/+bug/1444061 said a freeze, but that sounds like what you described up there14:27
ubot5Ubuntu bug 1444061 in mir (Ubuntu) "[regression] Mir servers freeze on startup (mouse and keyboard not responsive)" [Undecided,New]14:27
anpok_yes redid it by splitting the changes up into hm two iirc14:28
kdubanpok_, right, but how does splitting it up into two fix it?14:35
anpok_well .. i changed it along the lines of splitting it up :)14:35
anpok_the one that already landed siwtches the timeout handling to a timer fd (which is a new addition) and this change now ensure that there is an initial call to do the device scan14:36
kdubanpok_, ah, well, lgtm I suppose, maybe a comment in the bug about what the root cause was would be helpful though14:36
kdubanpok_, ah, yes, thats what I was asking about14:37
kdubanpok_,  +1 then from me14:38
=== dandrader is now known as dandrader|lunch
alan_gkdub: still need info? https://code.launchpad.net/~alan-griffiths/mir/more-surface-resize/+merge/25691216:35
=== essembe is now known as sbeattie
kdubalan_g, switched to abstain, probably enough for top-approval now16:37
kdubbut if not, i'll re-review later today16:37
alan_gkdub: I'm happy to leave it for later16:39
kdubalright, just in the thick of reorganizing some tests, will check later16:39
=== alan_g is now known as alan_g|EOD
=== dandrader|lunch is now known as dandrader
=== Guest34273 is now known as mibofra
=== dandrader is now known as dandrader|afk
=== dandrader|afk is now known as dandrader

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