mozilla :: #gfx

15 Mar 2017
06:39pulsebotCheck-in: - Ethan Lin - Bug 1346702 - Part1. Create border renderer in nsDisplayBorder::GetLayerState. r=mchang
06:39pulsebotCheck-in: - Ethan Lin - Bug 1346702 - Part2. Add CanCreateWebrenderCommands for checking if the render should create WR commands. r=mchang
07:28pulsebotCheck-in: - sotaro - Bug 1347050 - Remove unused workarond from ProcessWebRenderCommands r=nical
12:14pulsebotCheck-in: - sotaro - Bug 1347443 - Fix IsAccelAngleSupported() for RenderThread r=nical
13:42pulsebotCheck-in: - 213 changesets - Merge m-c to graphics
15:27kvarkllvm 4.0: "more aggressive aggressive dead code elimination" - I guess they are really into it :)
15:48lsalzmanthey made the ADCE more aggressive
15:56nicalWarning as errors in is tough
15:57nicaloften times it means you need to send a PR to the webrender repo, wait for review + landing, then merge it into the graphics branch before you get a chance to do what you want.
16:05lsalzmancan't disable it for just that module?
16:45digitaraldkats: any hypothesis why checkerboarding improved over the past beta versions:
16:55katsdigitarald: not really, no
18:09jrmuizelbotond: ping
18:23botondjrmuizel: pong
18:23jrmuizelbotond: where did you go?
18:23botondjrmuizel: food
18:23jrmuizelbotond: are you back now?
18:24botondjrmuizel: i am
18:24katsfood! that's what i forgot to do :p
18:24jrmuizelkats: good thing, otherwise I wouldn't have been able to get my r+
18:25kats(well, i didn't really forget. i decided to start setting up the new windows desktop and that is taking longer than i thought it would)
18:25jrmuizelkats: what did you end up buying?
18:27katsjrmuizel: the i7-7700
18:27jrmuizelkats: good
18:27katsi need to turn off as many windows services as i can though
18:27katsit's still quite slow
18:28katsalso maybe cloning over wifi isn't the best idea
18:28katsit's been going for a while
18:34kats... and it died
18:37botondjrmuizel: 2+ years, actually :/
18:37jrmuizelbotond: wow
18:45botondjrmuizel: although, it being called for every mouse event is considerably more recent than that
18:48katsjrmuizel: also i suspect stringifying the element is exceptionally expensive on that page
18:48botondyeah, that too
18:49jrmuizelkats: which Stringify are you referring to?
18:49botondjrmuizel: the Describe call
18:49katsit dumps the class attribute and id attribute and maybe a few other things
18:49jrmuizelbotond: isn't that what we removed?
18:49botondjrmuizel: yes. kats was saying is wasn't causing this much of a performance problem on most pages
18:50jrmuizelkats: ah, I misread
18:51jrmuizelkats: yes, I think that page is doing insane things
20:11nicaljrmuizel, kats: you around?
20:15jrmuizelnical: yes
20:32nicaljrmuizel: we are chatting with the security folks on vidyo (you were cc'ed on the emails, dunno if you are interested)
20:33jrmuizelnical: oh right
20:33jrmuizelnical: I
20:33jrmuizel'll join
20:38katsnical: are you muted? if not, can you try muting?
20:48nicalvidyo just crahed on me
21:09nicalI hate vidyo.
21:27mattwoodrowmstange: ping
21:27mstangemattwoodrow: pong
21:28mattwoodrowmstange: Got a few minutes to look at a layerization issue?
21:29mstangemattwoodrow: hmm... maybe
21:30mattwoodrowmstange: Ill post a bug comment and you can look when its convenient
21:30mstangemattwoodrow: that sounds good
21:30mstangethank you
21:47mstangemattwoodrow: whoooaaa
21:47mstangethat's exactly what I've always wanted
21:47mstangeI've always used which lets you share diffs by sending them as an email...
21:51mattwoodrowmstange: First time Ive used it, seems pretty good though
23:40mstangedvander: why is PGPU backwards?
23:40mstangedvander: GPUChild is in the parent process and GPUParent in the GPU process, right?
23:41dvandermstange: right. because child/parent is a blocking relationship in IPDL
23:41mstangedvander: ah!
23:41mstangethat makes sense
23:41dvandermatt got rid of The Sync Message though, if there are no more it could be flipped someday
23:42mstangeah ok
23:42mstangeI guess the sync messages that I was thinking of go through PAPZ
23:43dvanderoh yes, indeed. probably not gonna happen then
23:57dvandermstange: does the ViewID uniquely identify a framemetrics? like, if N layers have the same framemetrics, will those layers (and only those layers) have the same ViewID?
16 Mar 2017
No messages
Last message: 158 days and 14 hours ago