Issue metadata
Sign in to add a comment
|
FSLP positioning messed up for some sites |
||||||||||||||||||||||
Issue descriptionChrome Version: 64.0.3259.0 OS: macOS 10.12.6, 10.13.1 What steps will reproduce the problem? (1) Load the attached HTML file and follow the instructions. What is the expected result? The layout doesn't change when the video controls show and hide What happens instead? The video is contained in an element with overflow: hidden. The new FSLP path introduced in r513410 handles this poorly and positions the video — and its black superlayer — almost entirely offscreen.
,
Nov 14 2017
,
Nov 14 2017
,
Nov 14 2017
,
Nov 14 2017
Sorry chrishtr@, I wrote the original bug description as a quick note-to-self, it wasn't meant for human consumption :). Added a proper testcase and video.
,
Nov 20 2017
This is marked as Beta blocker and M-64 will be branched in ~1 week time. If possible, please plan the fix before branch point. Thank you!
,
Nov 20 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/a8745244573b42282c1edbb208b0442498a149c3 commit a8745244573b42282c1edbb208b0442498a149c3 Author: Sidney San Martín <sdy@chromium.org> Date: Mon Nov 20 19:03:55 2017 Use the regular layer tree for fullscreen low power detachment on Mac. Fixes an issue where fullscreen video wasn't positioned correctly in the low power tree for some sites, and a longstanding issue where clipped videos become un-clipped in the fullscreen low power tree. Made possible by two changes: - Black and transparent backgrounds are no longer rendered as IOSurfaces for color correction. For macOS to consider a layer solid black, it needs to actually have a black background color (and no contents). Note: I think we can avoid IOSurface replacement by setting a flag or two on the CAContext. See https://crrev.com/c/742663. - Instead of constructing a new layer tree for fullscreen low power, just make the root layer black and match its size to the background to meet the "opaque black superlayer" requirement for detachment. A nicer solution could be to send the browser a color, since it already has a solid color superlayer, in which case no special handling for fullscreen layer detachment would be needed at all, it would "just work". It still needs to be determined whether that would comply with the HTML fullscreen spec. See https://crbug.com/785001 . Bug: 781637 Change-Id: I94abb1ade9c58c218296a27f9a7bc3630d4a0c64 Reviewed-on: https://chromium-review.googlesource.com/775139 Commit-Queue: Sidney San Martín <sdy@chromium.org> Reviewed-by: ccameron <ccameron@chromium.org> Cr-Commit-Position: refs/heads/master@{#517879} [modify] https://crrev.com/a8745244573b42282c1edbb208b0442498a149c3/ui/accelerated_widget_mac/ca_renderer_layer_tree.h [modify] https://crrev.com/a8745244573b42282c1edbb208b0442498a149c3/ui/accelerated_widget_mac/ca_renderer_layer_tree.mm
,
Nov 21 2017
Verified the fix on the latest canary(64.0.3274.0) on Mac OS 10.12.6 using fslp_alignment.html and this seems to be working as intended. Hence adding the verified label.
,
Nov 22 2017
sdy@, please mark it as 'Fixed' if there is no other pending CL exists. Thanks..!
,
Nov 22 2017
|
|||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||
Comment 1 by chrishtr@chromium.org
, Nov 13 2017