1933288 - Add comprehensive WPT tests for 'stretch' keyword

Status: RESOLVED FIXED

(Core :: CSS Parsing and Computation, task)

Description

[Daniel Holbert [:dholbert]]

Filing this bug as a helper for bug 1789477 to land some tests for the CSS 'stretch' sizing keyword.

Comment 1

[Daniel Holbert [:dholbert]]

Attachment: Bug 1933288 part 1: Add a WPT to test the behavior of the 'stretch' keyword as an inline-size. r?dshin,#layout

Comment 2

[Daniel Holbert [:dholbert]]

Attachment: Bug 1933288 part 2: Add a WPT to test the behavior of the 'stretch' keyword as a block-size. r?dshin,#layout

I'm adding this test as a modified-copy of the similar inline-size test added
in the previous patch in this series.

Comment 3

[Daniel Holbert [:dholbert]]

Attachment: Bug 1933288 part 3: Add WPTs to test the behavior of the 'stretch' keyword as a min-inline-size and max-inline-size. r?dshin,#layout

The stretch value for these properties (when applied on a clamp on extreme
values) should mostly produce the same results as if we directly specified
"inline-size:stretch", so these tests have mostly the same expectations as the
equivalent test that just sets "inline-size".

Comment 4

[Daniel Holbert [:dholbert]]

Attachment: Bug 1933288 part 4: Add WPTs to test the behavior of the 'stretch' keyword as a min-block-size and max-block-size. r?dshin,#layout

The stretch value for these properties (when applied on a clamp on extreme
values) should mostly produce the same results as if we directly specified
"block-size:stretch", so these tests have mostly the same expectations as the
equivalent test that just sets "block-size".

Comment 5

[Daniel Holbert [:dholbert]]

Attachment: Bug 1933288 part 5: Add test expected-failure annotations. r?dshin,#layout

These new tests mostly fail for now (per these auto-generated annotations), but
they pass with my local patch stack for bug 1789477.

Comment 6

[Daniel Holbert [:dholbert]]

Notably, I'm not testing tables in these tests, because I found some interop differences between the behavior of stretch in tables, in my work-in-progress patch stack, vs. Blink, vs. WebKit (with -webkit-fill-available since they don't yet handle stretch).

It's not entirely clear what's correct for stretch in various table scenarios (since tables are underspecified and often involve at least one content-measuring pass where stretch would fall back to content-sizing; and many table-part CSS sizes are really just a starting suggestion; etc.), so I'm going to add those tests separately as .tentative or in the mozilla private WPT directory, since I have less confidence about what expectations are correct there.

Comment 7

[Daniel Holbert [:dholbert]]

Attachment: testcase for Chrome bug with min-block-size / max-block-size (i.e. min-height/max-height)

Comment 8

[Daniel Holbert [:dholbert]]

Attachment: testcase for Chrome bug with min-block-size / max-block-size (i.e. min-height/max-height)

Comment 9

[Daniel Holbert [:dholbert]]

Attachment: screenshot of buggy (I think) Chrome behavior on attached testcase

Comment 10

[Daniel Holbert [:dholbert]]

The testcase that I just attached (comment 8, screenshot in comment 9) is to demonstrate the bug (or at least inconsistency) that Chrome has with stretch for the min/max block sizing properties (typically min-height/max-height), which I alluded to in
https://phabricator.services.mozilla.com/D230159#7961971

I'll be filing a Chrome bug on that shortly, linking to the testcase/screenshot attached here.

Comment 11

[Daniel Holbert [:dholbert]]

Chrome bug for the min/max-block-size failures with nonzero inset values: https://issues.chromium.org/issues/380912396

Comment 12

[Daniel Holbert [:dholbert]]

Attachment: Bug 1933288 part 6: Add .tentative copies of the tests from previous patches, now testing "-webkit-fill-available", assumed to be an alias of "stretch". r?jwatt,#layout

These tests are all annotated as expected-failure for now, pending our
implementation of stretch/-webkit-fill-available landing and being enabled in
this directory in subsequent bugs.

Comment 13

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/6314926bcd51
part 1: Add a WPT to test the behavior of the 'stretch' keyword as an inline-size. r=jwatt
https://hg.mozilla.org/integration/autoland/rev/b0d3b4945584
part 2: Add a WPT to test the behavior of the 'stretch' keyword as a block-size. r=jwatt
https://hg.mozilla.org/integration/autoland/rev/de4f982de09c
part 3: Add WPTs to test the behavior of the 'stretch' keyword as a min-inline-size and max-inline-size. r=jwatt
https://hg.mozilla.org/integration/autoland/rev/f78271585a62
part 4: Add WPTs to test the behavior of the 'stretch' keyword as a min-block-size and max-block-size. r=jwatt
https://hg.mozilla.org/integration/autoland/rev/8f90d4080eec
part 5: Add test expected-failure annotations. r=jwatt

Comment 14

[Daniel Holbert [:dholbert]]

Try's looking good so far for parts 1-5 ( https://treeherder.mozilla.org/jobs?repo=try&revision=1af82ac5e8970e727fe28ad50f9a6f80f1106b22 ), so I landed those.

--> leave-open to land part 6 (which just adds copies with -webkit-fill-available) once that's r+.

Comment 15

[Web Platform Test Sync Bot [:wpt-sync] (Matrix: #interop:mozilla.org)]

Created web-platform-tests PR https://github.com/web-platform-tests/wpt/pull/49364 for changes under testing/web-platform/tests

Comment 16

[pstanciu]

https://hg.mozilla.org/mozilla-central/rev/6314926bcd51
https://hg.mozilla.org/mozilla-central/rev/b0d3b4945584
https://hg.mozilla.org/mozilla-central/rev/de4f982de09c
https://hg.mozilla.org/mozilla-central/rev/f78271585a62
https://hg.mozilla.org/mozilla-central/rev/8f90d4080eec

Comment 17

[Daniel Holbert [:dholbert]]

(In reply to Daniel Holbert [:dholbert] from comment #14)

--> leave-open to land part 6 (which just adds copies with -webkit-fill-available) once that's r+.

whoops, I forgot to add leave-open. Reopening, though I just triggered lando for part 6 and we can let this close when that makes it to m-c.

Comment 18

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/8c4ccbbd99e2
part 6: Add .tentative copies of the tests from previous patches, now testing "-webkit-fill-available", assumed to be an alias of "stretch". r=jwatt

Comment 19

[Daniel Holbert [:dholbert]]

I'll be adding another patch here to extend these tests, actually. --> leave-open after all.

Comment 20

[Daniel Holbert [:dholbert]]

Attachment: Bug 1933288 part 7: Add a second pass to 'stretch' WPTs to test with "box-sizing:border-box", to validate that it doesn't impact the sizing. r?jwatt,#layout

box-sizing:border-box shouldn't impact the sizing because we're growing to fill
the containing block, which means we end up with the same overall size in the
relevant axis regardless of whether we're stretching the border-box or the
content-box.

(There's one special case in these tests where it does actually matter, but
not in a way that's really based on how 'stretch' itself is resolved; so I
simply opt that subtest out of this second pass.)

Comment 21

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/7892bfc41933
part 7: Add a second pass to 'stretch' WPTs to test with "box-sizing:border-box", to validate that it doesn't impact the sizing. r=TYLin

Comment 22

[agoloman]

https://hg.mozilla.org/mozilla-central/rev/8c4ccbbd99e2
https://hg.mozilla.org/mozilla-central/rev/7892bfc41933

Comment 23

[Web Platform Test Sync Bot [:wpt-sync] (Matrix: #interop:mozilla.org)]

Upstream PR merged by moz-wptsync-bot