880104 - Remove -moz prefixed style from flexbox tests, to make them more suitable for w3c testsuite & other browser vendors to run

Status: RESOLVED FIXED

(Core :: Layout, defect)

Description

[Daniel Holbert [:dholbert]]

Filing this bug on removing some -moz prefixed styles (e.g. "-moz-max-content") from our flexbox reftests, so that other browsers have a shot at rendering them and so that they're more suitable for submission to the w3c.

Comment 1

[Daniel Holbert [:dholbert]]

Attachment: part 1: Rewrite flexbox-align-self-*

The flexbox-align-self-* reftests (reference cases in particular) use intrinsic sizing keywords a lot, to shrinkwrap chunks of text.

This patch tweaks these tests to use floats and/or inline-blocks instead of intrinsic sizing keywords (and does a bit of other cleanup while I'm at it).

Comment 2

[Daniel Holbert [:dholbert]]

Attachment: part 2: replace some -moz prefixed sizing keywords with their already-known exact pixel values

Comment 3

[Daniel Holbert [:dholbert]]

Attachment: part 3: fix flexbox-justify-content-horiz-5-ref.xhtml to not rely on -moz-max-content

This patch uses the same text-align:center + inline-block technique used in part 1, for centering shrinkwrapped content, now in flexbox-justify-content-horiz-5-ref.xhtml. (plus some other cleanup / simplification in that test)

Comment 4

[Daniel Holbert [:dholbert]]

Attachment: part 4: remove unnecessary intrinsic sizing keyword from flexbox-widget-flex-items-4.html

This part removes an unnecessary (on my machine at least) intrinsic sizing keyword from flexbox-widget-flex-items-4.html, and does a bit of other minor cleanup there (making the testcase and reference case align slightly better, property-ordering-wise).

Comment 5

[Daniel Holbert [:dholbert]]

Try run: https://tbpl.mozilla.org/?tree=Try&rev=03896438777e

dbaron rubber-stamped the general idea of this bug, so assuming the Try run succeeds, I'll go ahead and land this.

Comment 6

[Daniel Holbert [:dholbert]]

(In reply to Daniel Holbert [:dholbert] from comment #5)
> dbaron rubber-stamped the general idea of this bug,

(in IRC)

Comment 7

[Daniel Holbert [:dholbert]]

Try run in comment 5 looks good, so I pushed:
  https://hg.mozilla.org/integration/mozilla-inbound/rev/5e74b815c332
  https://hg.mozilla.org/integration/mozilla-inbound/rev/1e838ba3ecb5
  https://hg.mozilla.org/integration/mozilla-inbound/rev/9412ec5c0b65
  https://hg.mozilla.org/integration/mozilla-inbound/rev/5c70ec981c18

Comment 8

[Daniel Holbert [:dholbert]]

Apparently the simplification to flexbox-align-self-vert-rtl-1-ref.xhtml made it fail on Android, with a single letter in "unspec" being off by one pixel (or maybe less) -- the middle "s". Sample log:
 https://tbpl.mozilla.org/php/getParsedLog.php?id=23849919&tree=Mozilla-Inbound

I just pushed a followup to annotate that test as having 32 fuzzy pixels on Android.
  https://hg.mozilla.org/integration/mozilla-inbound/rev/0972920ff3be

Comment 9

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/5e74b815c332
https://hg.mozilla.org/mozilla-central/rev/1e838ba3ecb5
https://hg.mozilla.org/mozilla-central/rev/9412ec5c0b65
https://hg.mozilla.org/mozilla-central/rev/5c70ec981c18
https://hg.mozilla.org/mozilla-central/rev/0972920ff3be

Comment 10

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

(In reply to Daniel Holbert [:dholbert] (traveling June 7-12, w/ mixed availability) from comment #8)
> Apparently the simplification to flexbox-align-self-vert-rtl-1-ref.xhtml
> made it fail on Android, with a single letter in "unspec" being off by one
> pixel (or maybe less) -- the middle "s". Sample log:
>  https://tbpl.mozilla.org/php/getParsedLog.php?id=23849919&tree=Mozilla-
> Inbound
> 
> I just pushed a followup to annotate that test as having 32 fuzzy pixels on
> Android.
>   https://hg.mozilla.org/integration/mozilla-inbound/rev/0972920ff3be

That makes it seem like the centering methods used in test and reference round differently.  It might be worth investigating.