706194 - Adapt and check in tests from webkit for new unicode-bidi attributes

Status: RESOLVED FIXED

Description

[Simon Montagu :smontagu]

Ryosuke Niwa from webkit.org suggested that we share tests for unicode-bidi: isolate and plaintext between Gecko and Webkit.

Here are WebKit tests:
http://trac.webkit.org/browser/trunk/LayoutTests/fast/text/international/unicode-bidi-plaintext.html
http://trac.webkit.org/browser/trunk/LayoutTests/fast/text/international/unicode-bidi-plaintext-in-textarea.html
http://trac.webkit.org/browser/trunk/LayoutTests/css3/unicode-bidi-isolate-basic.html
http://trac.webkit.org/browser/trunk/LayoutTests/css3/unicode-bidi-isolate-aharon.html
http://trac.webkit.org/browser/trunk/LayoutTests/css3/bdi-element.html
http://trac.webkit.org/browser/trunk/LayoutTests/css3/unicode-bidi-insolate-parse.html
http://trac.webkit.org/browser/trunk/LayoutTests/fast/css/default-bidi-css-rules.html

Comment 1

[Simon Montagu :smontagu]

Created attachment 584392 [details] [diff] [review]
Patch part 1

For clarity I'll do this in 2 parts: this is the initial adaptation of the webkit tests to our test harnesses, which has lots of failures.

Comment 2

[Simon Montagu :smontagu]

Created attachment 584394 [details] [diff] [review]
Patch part 2

... and these are the changes to make the tests pass (modulo bugs noted in comments)

Comment 3

[Simon Montagu :smontagu]

Comment on attachment 584394 [details] [diff] [review]
Patch part 2

Reasons for the changes:

>+    text-align: left;
 in a number of tests, because of a difference between the way that we and webkit handle unicode-bidi: plaintext. If no text-align is specified, we align paragraphs to the left and right depending on the resolved direction of each paragraph, but webkit aligns everything depending on the direction of the enclosing element. See http://lists.w3.org/Archives/Public/www-style/2011Oct/0846.html and https://bugs.webkit.org/show_bug.cgi?id=71194

>diff --git a/layout/reftests/bidi/unicode-bidi-isolate-basic.js b/layout/reftests/bidi/unicode-bidi-isolate-basic.js

>-    var strongRTLs = ['ש', 'נ', 'ב', 'ג', 'ק', 'כ', 'ע'];
>+    var strongRTLs = ['א', 'ב', 'ג', 'ד', 'ה', 'ו', 'ז'];

Using Hebrew letters in alphabetical order makes the test results easier to read for me (the original used the same key positions as a,b,c,d,e,f,g)

>diff --git a/layout/style/test/test_default_bidi_css.html b/layout/style/test/test_default_bidi_css.html

>-    ['div', {'dir': 'rtl'}, 'rtl', 'embed'],
>+    ['div', {'dir': 'ltr'}, 'ltr', '-moz-isolate'],
>+    ['div', {'dir': 'rtl'}, 'rtl', '-moz-isolate'],
>     ['div', {'dir': 'auto'}, 'ltr', '-moz-isolate'],
>-    ['div', {'dir': ''}, 'ltr', 'embed'],
>+    ['div', {'dir': ''}, 'ltr', '-moz-isolate'],

>-    ['pre', {}, 'ltr', 'normal'],
>-    ['pre', {'dir': 'ltr'}, 'ltr', 'embed'],
>-    ['pre', {'dir': 'rtl'}, 'rtl', 'embed'],
>+    ['pre', {}, 'ltr', '-moz-isolate'],
>+    ['pre', {'dir': 'ltr'}, 'ltr', '-moz-isolate'],
>+    ['pre', {'dir': 'rtl'}, 'rtl', '-moz-isolate'],
>     ['pre', {'dir': 'auto'}, 'ltr', '-moz-plaintext'],
>-    ['pre', {'dir': ''}, 'ltr', 'embed'],
>+    ['pre', {'dir': ''}, 'ltr', '-moz-isolate'],

Another difference between us and webkit: we give block elements with a dir attribute -moz-isolate, not embed (https://www.w3.org/Bugs/Public/show_bug.cgi?id=14850 and https://bugs.webkit.org/show_bug.cgi?id=65617)

Comment 4

[Simon Montagu :smontagu]

Created attachment 603424 [details] [diff] [review]
Patch part 3

Assuming that we want to WONTFIX bug 713621, this makes the necessary changes to test_default_bidi_css.html

Comment 5

[David Baron :dbaron: ⌚️UTC-7]

Comment on attachment 603424 [details] [diff] [review]
Patch part 3

r=dbaron

Comment 6

[Simon Montagu :smontagu]

https://hg.mozilla.org/integration/mozilla-inbound/rev/d8d124d2a76c
https://hg.mozilla.org/integration/mozilla-inbound/rev/af0733d23b87
https://hg.mozilla.org/integration/mozilla-inbound/rev/ef43fa1005b7

Comment 7

[Ed Morley [:emorley]]

Sorry, backed out (https://hg.mozilla.org/integration/mozilla-inbound/rev/c5ead08145a9) for what looked like bustage, but was slightly ahead of myself (and clearly still half asleep!) as was just hg issues. 

Relanded:
https://hg.mozilla.org/integration/mozilla-inbound/rev/75ec3a324eee
https://hg.mozilla.org/integration/mozilla-inbound/rev/386b8a429bbf
https://hg.mozilla.org/integration/mozilla-inbound/rev/3af815def682

Comment 8

[Ed Morley [:emorley]]

Followup to fix bad qimport where the non-ASCII text in the tests got mangled :-(

https://hg.mozilla.org/integration/mozilla-inbound/rev/697b229db32a

(Having a successful morning here!)

Comment 9

[:Ehsan Akhgari (needinfo please, extremely long backlog)]

https://hg.mozilla.org/mozilla-central/rev/75ec3a324eee
https://hg.mozilla.org/mozilla-central/rev/386b8a429bbf
https://hg.mozilla.org/mozilla-central/rev/3af815def682
https://hg.mozilla.org/mozilla-central/rev/697b229db32a

Comment 10

[:Ehsan Akhgari (needinfo please, extremely long backlog)]

https://hg.mozilla.org/mozilla-central/rev/b445a3255696
https://hg.mozilla.org/mozilla-central/rev/662c2b50b803
https://hg.mozilla.org/mozilla-central/rev/93b1912f077f