1272181 - [writing-mode] Inconsistent float layout with "writing-mode: vertical-rl" and @font-face between Reload and Shift+Reload

Status: NEW

(Core :: Layout: Block and Inline, defect)

Description

[Mats Palmgren (inactive)]

Attachment: Ahem.ttf

STR:
1. Shift+Reload the attached testcase
2. Reload the attached testcase

ACTUAL RESULTS
There is a difference in layout after 1 compared to after 2.

EXPECTED RESULTS
No difference.

Comment 1

[Mats Palmgren (inactive)]

Attachment: testcase

Comment 2

[Mats Palmgren (inactive)]

Attachment: testcase

Comment 3

[Mats Palmgren (inactive)]

Bah, it seems the test doesn't work as a Bugzilla attachment...
Please save the two files and update the @font-face url accordingly.

Comment 4

[Gérard Talbot]

Mats,

I am able to reproduce the problem locally (but erratically and inconsistently) and also with a reduced version of your test.

Note that:

- writing-mode no longer requires the -webkit- prefix:
https://bugs.chromium.org/p/chromium/issues/detail?id=533448
- when an inline is has a different writing-mode value than its containing block, then its computed display is 'inline-block':
https://drafts.csswg.org/css-writing-modes-3/#block-flow
- if line-height: 100%; for all elements, then I do not see what 'justify-self:baseline' does in the test
- I have reduced your test to a 5 declarations test

Chances are very high that your test is a duplicate of 
http://test.csswg.org/suites/css-writing-modes-3_dev/nightly-unstable/html/orthogonal-parent-shrink-to-fit-001j.htm
or some other combinations of orthogonal-parent-shrink-to-fit-001* available in
http://test.csswg.org/suites/css-writing-modes-3_dev/nightly-unstable/html/chapter-7.htm#s7.3
which we cover in bug 1166120 . So chances are this is a DUPLICATE of bug 1166120 unless there is a Reload and Shift+Reload issue.

Comment 5

[Mats Palmgren (inactive)]

Yeah, the testcase can probably be reduced more.  It was derived from a baseline
alignment test I wrote for grid while working on support for those properties
there.  We don't support them at all in block layout (yet).

> unless there is a Reload and Shift+Reload issue.

There is a difference in layout between those two, which ultimately what this bug
is about.  When a @font-face load is completed we basically re-layout everything
and it appears the type of Reload makes difference in which order we do things
which leads to the error.

I haven't looked at bug 1166120 in detail, but judging from the summary there's
a chance it might be the same underlying bug.  Marking dependent for now.