Issue metadata
Sign in to add a comment
|
requestIdleCallback should not require threaded compositing in layout tests |
||||||||||||||||||||||||
Issue descriptionIt appears that requestIdleCallback requires threaded compositing in layout tests to work properly, which is a landmine for developers who want to use this feature when implementing their own layout tests. It seems naively like you could just pretend like there's an idle period after every raf and just make it work. See also: https://groups.google.com/a/chromium.org/forum/#!msg/graphics-dev/f3vqLzm22c4/4daCc83gBQAJ.
,
Mar 17 2016
Doing it after every raf isn't a good solution - the raf is at a different point in the rendering pipeline, so arguably this would be testing something which is even more different than never running the idle callbacks at all (which is actually a valid situation if the page is busy). The solution is to hook up the non-threaded compositor to provide the necessary signals to the scheduler when it draws frames. Sami thought that this should just work, but there must be a bug somewhere which is preventing these signals getting through. I won't have the time (or the expertise in the compositor) to do this in the next couple of weeks so making this bug available if anyone else gets a chance to look into it.
,
Mar 21 2016
I don't *think* there's anything stopping us from using real idle periods in the single threaded proxy. Let me dupe against the other bug since both are really about the same thing.
,
May 18 2016
|
|||||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||||
Comment 1 by szager@chromium.org
, Mar 16 2016