Join GitHub today
GitHub is home to over 50 million developers working together to host and review code, manage projects, and build software together.
Sign upBetter handling of iframes in the compositor tree #3197
Conversation
hoppipolla-critic-bot
commented
Sep 2, 2014
|
Critic review: https://critic.hoppipolla.co.uk/r/2495 This is an external review system which you may optionally use for the code review of your pull request. In order to help critic track your changes, please do not make in-place history rewrites (e.g. via |
|
Note that I did not include Bryan's work-around for the failing test in my version, hoping that the reworked tests and code aren't flaky. If travis fails, I'll look into fixing it. |
|
|
Yeah, I'll have to poke at this some more. I tried replicating Bryan's original work-around (versus my reworked one) and this is still failing. |
mrobinson
force-pushed the
mrobinson:iframes
branch
2 times, most recently
from
ec6107f
to
a2c0cec
|
Wow! I have to say I'm completely baffled by the continual failure of this reftest. In the latest round, I removed the content of the inner frames, which should make it resilient to iframe content clicking in later and what Bryan's reftest looked like. Still the test fails though. |
mrobinson
force-pushed the
mrobinson:iframes
branch
3 times, most recently
from
26bbd51
to
b04eff8
|
I modified the failing reference test to eliminate all chances of flakiness, but still allow verifying the fix for the old problem of calling SetLayerClipRect on a non-existent layer. I think the remaining failures are unrelated to this changes in this branch. |
|
I've retriggered the wpt2 job a lot of times now, and I'm pretty sure it's consistently failing on Document-characterSet-normalization.html. Could you run that locally and make sure it passes for you? |
|
When I run check-wpt locally, /html/dom/interfaces.html fails, but not characterSet-normalization.html. I'll keep investigating. |
|
We're disabling the normalization test on master now, so you can ignore it. |
Instead of waiting to create the root layer, create them as soon as the Compositor receives the frame tree. This allows the compositor to create a layer tree skeleton of which to hang the base layers when they are ready.
The constallation has accurate information about iframe layer origins, but not their size.
Instead of producing image output as soon as the first pipeline is ready, we wait to produce the output until all pipelines are in the idle RenderState. This should remove a race condition when running reference tests.
Add iframe tests to their own subdirectory and add another test case that used to trigger a fatal error. The new test case uses the "allow-scripts" sandbox attribute to work around a script task failure caused by the child frame sharing the same script task as the parent.
Better handling of iframes in the compositor tree
mrobinson
merged commit 443bcc4
into
servo:master
mrobinson commentedSep 2, 2014
This is a rework of Bryan Bell's original PR from #2943. It splits the single commit into independent pieces, preserves the existing "root layer + base layer" compositor tree layout, and includes miscellaneous other cleanup.