mozilla :: #gfx

16 Mar 2017
00:02mstangedvander: mostly, with one exception: http://searchfox.org/mozilla-central/source/gfx/layers/FrameMetrics.h#934-935
00:03mstangedvander: your second question seems a bit nebulous to me - "the same framemetrics" could mean "all fields of the two frame metrics are the same, including the ViewID" and that trivially makes it true
00:04mstangeso I answered the reverse question - if two FrameMetrics have the same ViewID, are the rest of the fields also guaranteed to be the same? And the answer to that is "almost".
00:04dvandermstange: ah I see. so, that is true of ancestor masks as well?
00:05mstangedvander: yes
00:05mstangethat should probably be clarified somewhere
00:05mstangegood call
00:05dvanderactually you just said scroll clip can be different
00:05mstangeoh
00:05dvanderdoes that only apply to non-mask clips?
00:06mstangeI forgot that the API there is actually quite nice now
00:06mstangeso
00:06mstangethe exception applies to all types of clips, rectangular and masks
00:06mstangeso the comment is actually in the right place
00:07dvanderah okay. thanks
08:55pulsebotCheck-in: https://hg.mozilla.org/projects/graphics/rev/9cf2ea8fb1b7 - Ethan Lin - Bug 1347445 - Create border layer for non-QR case. r=mchang
15:01kvarknical: you here?
16:26nicalkvark I am
16:28kvarknical: we were just wondering about where everyone disappeared during the call :)
16:28nicalkvark yeah sorry I realized a bit late that there are a couple of important vaccines I need before going to Thailand
16:28kvarkoh sweet
16:30kvarkI guess you'll be off during the 1H of April
16:31nicalI'll be off half of April
16:31nicalAnd the good thing is, if I get bitten by a dog I won't die (probably).
16:32nicalI just commented on the object destruction issue
16:32nicalI think that you can't specify a behavior that depends on the state of the GC
16:33nicalthe timing of the destruction of garbage collected things is by spec undeterministic (and unobservable)
16:34nicalI am not sure if it contradicts or not something you've written in the proposal, but that has to be taken into account
16:35kvarkwe aren't trying to specify the GC-related behavior. Regardless of GC, we can check the objects that are no longer references by the client, and remove their backing resources (in `cleanup()`)
16:41kvarknical: want to see something cute? look at the other PR ;)
16:43nicalyeah I wasn't sure what you meant by out of scope
16:44nicalif it meant, after GC or as soon as not referenced
16:44* nical goes to look for something cute
16:45nicalcute indeed
16:46nicalalthough, their proposals are at a very early stage (or at least apple's not sure about google's) so I am not sure that it's fair to do side by side comparisons at this point
16:46kats*sigh* i think i bricked my spare router trying to set it up as a wireless repeater
16:47katsthis is not turning out to be a great week
16:47* lsalzman is trying to escape while he still can, currently packing for PTO.
16:48lsalzmani suppose that is no guarantee the week still can't get bad for me, even while away
16:49kvarknical: true, but I think the general direction is pretty settled by now. It's not meant to show the end result of the proposals, since I assume we'll all converge to one or the other. It's meant to show what we have now.
16:54nicalOk I haven't followed apple's proposal closely but they seemed to be expecting to progressively move away from metal as the spec gets fleshed out in their initial annoucement.
16:55nicalIn any case I don't think it's harmful to have that especially if it isn't shoved on their face in the middle of the proposal document
16:56kvarknical: yes, that's correct. I'm not sure why they'd want to move away from Metal though. At least it's somewhat familiar :)
16:58nicalwell it sounded to me like "we started with metal as an initial prototype because it was easy for us but we expect that it's going to be more painful for everybody else and that they'll push for something else and that's fine."
16:59kvarksounds so Apple, right? /irony
17:00nicalthat new API aint gonna have a jack port, no no no.
17:01nicalI guess a lot of these things will be clearer after the first face to face meetings
17:02nical"target audience: | human beings | human beings | over-caffeinated engine developers
17:04kvarklol, pretty much :)
17:10rhuntkvark: is the 3d api work public? and if so where can i look at it?
17:11kvarkrhunt: it's in a private repo, but I can share the draft with you
18:39rhuntmchang: ping
18:40mchangrhunt: pong
18:40rhuntyou still getting crashes when you click on the hamburger menu?
18:40mchangrhunt: umm i was 1-2 days ago, i haven't updated to today's master
18:40mchangrhunt: did you fix it?
18:41rhuntmchang: no, im seeing if i can repro it
18:41rhuntso far no
18:41mchangrhunt: this is with WR enabled?
18:41rhuntmchang: yes
18:41mchangrhunt: hmm odd, on today's nightly?
18:42rhuntmchang: yeah i updated to latest today, haven't tried before
18:42rhuntmchang: running with --enable-optimize --enable-debug --enable-webrender
18:42mchangrhunt: alright let me try a new build
18:43rhuntmchang: okay
20:39jyadvander: Bas said you would be interested with this one
20:40jyaassertion on start on this brand new dell XPS 15 (with intel kaby-lake and nvidia 1050)
20:40jyahttps://irccloud.mozilla.com/pastebin/1CWusSUq/
21:42digitaraldkats: for FT and our discussion, are we getting closer aligned or moving away from each other?
21:42digitaraldwe can jump on a call if that helps to dig through the remaining issues
21:55dvanderjya: that is super weird. i think jrmuizel knows more about that condition than I, it's based on registry settings or something
22:12jyadvander: I've tried installing the latest intel drivers from intel web site. But this doesn't help
17 Mar 2017
No messages
   
Last message: 9 days and 13 hours ago