Layout Team Weekly Snippets - October 9, 2017

[Emil A Eklund]

Please use the following document for Q4 OKR planning:
https://docs.google.com/document/d/1ci6s93m4r0bcuGHkVbyZ_VnVoOsUJsRjggwnZvbfuO8/edit
You *should* all have access to it already, if not click the
"request permission" button and I'll grant it.

Updates since October 2:
- Refactored LayoutTheme by breaking all dependencies on LayoutObject
to allow it to be shared across legacy layout and LayoutNG. Still
some work to be done to support Shadow DOM children of some of the
decorated controls.
- Started working on hit testing for LayoutNG with fragment painting
enabled as the existing hit testing implementation requires the
legacy layout tree to be populated, which we no longer do when
fragment painting is enabled.
- Last week the number of open Blink>Layout bugs dropped below 800 for
the first time (down from 1000 earlier this year). Thanks for all
your help in triaging, fixing, and closing issues!
- Q4 planning.

This week:
- Continue working on hit testing.
- More Q4 planning.

Please reply-all with your own update for last week and also indicate
if any of the proposed topics sounds interesting to you or if you have
another one you'd like to add to the agenda.

[Aleks Totic]

ResizeObserver

- still stuck at a single LGTM.

- there is a discussion among smart people whether ResizeObserver should be an event instead. Not something I like to see when I am ready to ship. We should do the right thing for webdev, whatever it is.

ResizeObserver: instrument and analyze current usage on top polyfill sites. Emailed results to blink-dev. I am now comfortable with chrome://tracing, feel free to ask for help.

ResizeObserver: remove duplicate local tests after tests have moved to w-p-t 702676

LayoutNG

- fixed nested NG OOF blocks inside legacy container 704015

Etc

- Fix gdb WTF::String printer 702816. [impl_][ptr_] became [impl_][ptr_][ptr_]. Do our strings now have an additional degree of indirection? Do templates optimize it away? No idea, but my gdb works again.

Aleks

[Philip Rogers]

Hi layout friends, I normally hang out in paint/ but I'll be helping finish Root Layer Scrolling (RLS) this quarter with Steve (skobes@chromium), Stefan (szager@chromium), David (bokan@chromium), and Suneel (suneel.kota@samsung).

Last week:

- In CQ: add "attempt_to_remove_scrollbars" for root layer scrolling which fixes 89 layout test failures.

- Landed 4 patches cleaning up unit tests for RLS; no non-NG unit tests fail with root layer scrolling now.

- Administrivia: Cleaned up the RLS bug burndown list, setup standing meeting, kicked off RLS trybot setup (https://crbug.com/771753).

- Root layer scrolling reviews (Stefan's popup fix, Steve's expectations updates).

- Updated linux and mac build docs with jumbo build instructions.

- Discussion around async image decoding as well as how to gather statistics for caching decodes across navigations (https://crbug.com/769051).

[Steve Kobes]

Thanks Philip, we're excited to have you helping out. :)

[RLS]

- landed try-flag v0 (crrev.com/507206) - downloads JSON from finished trybots and prints unexpected fail / pass with merged platform specifiers

- working on try-flag v1 - create _tryflag branch, write rwt.flag, upload and trigger

- regenerate flag expectations using try-flag v0 (crrev.com/507000)

- assisting sataya.m with iframe border bug (crbug.com/746777)

- discuss PLSA scrollbar add/remove strategy with pdr

- assist szager with transform origin investigation

- cleanup and manage outstanding bug spreadsheet (bit.ly/rls-bugs)

[other]

- meet with erikchen about display:none iframes, did some research on DisplayNoneIFrameCreatesNoLayoutObject and the Dispose() problem (summarized in crbug.com/771757)

- preparing "Life of a Pixel"

[Morten Stenshorne]

Recently:

LayoutNG: Multicol min/max width calculation, and initial support for multicol balancing

Also wasted a day on compilation problems. Turned out that it was caused by enabling the lld linker, which supposedly is a great linker for goma, but it broke down completely with icecc, because it cannot use bundled binutils.

Next up:

LayoutNG: Not sure what. I shouldn't start on anything big, since I'm going away for a while soon. But maybe:

A: Improve column balancing.

B: Improve fragmentation to support line boxes, floats and replaced content.

C: Figure out how to deal with fragmentation when entering legacy layout and re-entering NG. Seems tricky. Seems like we'd kind of need legacy layout to abort at fragmentainer boundaries, return to NG and resume legacy layout in the next fragmentainer or something. So we'd have to add some NG awareness to all the legacy layout modes, such as tables, flexbox and grid? Doesn't sound good.

Right now I'm thinking B.

--
You received this message because you are subscribed to the Google Groups "layout-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to layout-dev+unsubscribe@chromium.org.
To post to this group, send email to layou...@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/layout-dev/CAAjAU%3DaipG1abcv%3D1M_HjuO7QoixUHBNOA9aauGD8niSapcBrg%40mail.gmail.com.

[Koji Ishii]

Updates since October 2:

- Implemented list-item 'outside' positioning. Symbols still use

  incorrect fonts, and image not supported yet.

- Implemented 'text-overflow: ellipsis' (only for NGPaint.)

- Fragment-based baseline for pre-NG ancestors.

- Fix dynamic renumbering of list-items.

- Fix inline blocks with orthogonal writing modes.

- Some cleanup and additions to geometry classes.

- Continue working on VisualRect for fragments.

This week:

- Continue working on VisualRect for fragments.

- Investigate -webkit-rtl-ordering.

- Investigate test failures.

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/layout-dev/CAJpGdU-uqPb3zGp-gJNG5Z8bRzLBmPnZbpU8Nmp%3Dy2VNQRTqjA%40mail.gmail.com.

[Ian Kilpatrick]

I've been working on a series of changes to make NGInlineLayoutAlgorithm to produce fragments in a linebox-by-linebox fashion. Nearly there!

(this will allow us to fragment properly for linebox continuations + simplify some of the linebreaker code, etc).

https://chromium-review.googlesource.com/c/chromium/src/+/703512

https://chromium-review.googlesource.com/c/chromium/src/+/706343

https://chromium-review.googlesource.com/c/chromium/src/+/709716

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/layout-dev/CACQRE%2BTeG99j9foVPqmk7MEFifksBDzZerwKsNp5KO1JTVjp3A%40mail.gmail.com.

[Karl Anders Øygard]

Time off most of last week, so not a lot of activity.

- Landed fix for the overpainting bug (https://crrev.com/c/695109)

- Discussed the NG table cell patch (https://crrev.com/c/649698) with Morten.  Reviving it now, and will try to rework it into something hopefully a little more palatable.

Karl

[Xiaocheng Hu]

Last week:

- Gave up the idea of TextIterator-free editing utilities

- Ended up studying editing functions using InlineBox, and designing new LayoutNG APIs for them

This week:

- Rewrite position canonicalization on top of offset mapping

- Prototype another mapping from inline node text content to line structures

--
You received this message because you are subscribed to the Google Groups "layout-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to layout-dev+unsubscribe@chromium.org.
To post to this group, send email to layou...@chromium.org.

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/layout-dev/CAGN-iCQL7ShamrX0zNFTFwG9WA9cR-0bhdugQGQ7mbKLg%2BR_9w%40mail.gmail.com.

[Dominik Röttsches]

Last week,

• Prepared spec changes and test case to nudge text-decoration-skip: ink objects; discussion

• Implemented CBCL/CBDT, CFF2 and SBIX instantiation + testing

• Reviewed Freetype + HarfBuzz joint GN target CL

This week in a workshop with Behdad, addressing Mac trak table system font letter spacing, and other things.

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/layout-dev/CAFqEGhY-rXzHApnioem9YN8oRbMmG9aYWY6182N_qunN27KYiQ%40mail.gmail.com.

[Dominik Röttsches]

PS: Landed a fuzzer for HarfBuzzShaper as part of Fuzzathon.