Monday, April 27, 2015
Performance Tracking (benjhayden)
- Preliminary LayoutAnalyzer-powered UMA findings. A few surprises
already, lots of opportunity for further analysis and more data.
- Working on LayoutTree trace object snapshots and various
LayoutAnalyzer improvements.
Scrolling (skobes) [crbug.com/417782]
- Reviewed bokan's RootFrameViewport change, and investigated a
background painting issue with root layer scrolling + slimming paint
(trchen is now working on a fix).
- Plan to work on testing root layer scrolling on Android this week.
Looking at failing layout tests related to fractional scroll offsets.
Line Boxes (szager) [crbug.com/321237]
- Continued work on line layout LayoutUnit conversions. Have a patch
that sets the barrier between float and LayoutUnit at the InlineBox
level, which seems like the sensible place.
- There is still a performance regression that I haven't been able to
eliminate. Was able to get some small performance wins by looking at
profiling data, but I've hit a wall with that approach. Will discuss
next steps and come up with a plan this week.
Flexbox (cbiesinger) [crbug.com/426898]
-
More flexbox updates.
-
Trying to get answers to flexbox spec questions, was promised answers
this week.
List marker refactoring (dsinclair) [crbug.com/370461]
- Working on performance of list marker code. New version is slower
then old version. html5-full-render on ToT was ~16.5 seconds for me,
with ListMarkerPseudoElement is ~17.5 seconds.
- Discussions this week on how to proceed with ListMarkerPseudoElement.
Misc Warden (dsinclair, pilgrim)
- Finishing work on decoratedName split, all annotations are moved
over, just finishing off rebaselines. (dsinclair)
-
More renderer to layoutObject renames. (dsinclair)
-
Wrapped up selection gap clipping rework. (wkorman)
-
Working on slimming LayoutObject and pushing bits down to
descendants. (pilgrim)
Text (eae)
- Added unit tests for complex text shaping (eae)
Style resolution (rune)
- Working on type selectors for SVG in HTML documents and picking up
componentized style resolver work again. There is some question
about whether the SVG bug is worth fixing.
Region-based multi-column support (mstensho) [crbug.com/334335]
- Found a new crash. Maybe something changed on trunk. Will fix then
reenable multi-col.
Layout refactoring
- esprehn, ikilpatrick, cbiesinger, ojan, leviw working on coming up
with concrete proposals.