Upstream changes made to rotate CSS property parsing test in WebKit r276554
#28691
Conversation
There was a problem hiding this comment.
This mostly looks good, modulo the comment below... which I guess is only partly related, so I'm ok with landing this and fixing that in a followup as well if you prefer.
Though I wonder if this was reviewed as part of WebKit's review process, and, if so, whether it should be auto-imported without review in the WPT repo, like such changes made through the Mozilla or Chromium review processes.
| test_valid_value("rotate", "0 1 0 400grad", "y 400grad"); | ||
|
|
||
| test_valid_value("rotate", "z 400grad"); | ||
| test_valid_value("rotate", "400grad z", "z 400grad"); | ||
| test_valid_value("rotate", "0 0 0.5 400grad", "z 400grad"); | ||
| test_valid_value("rotate", "0 0 1 400grad", "z 400grad"); |
There was a problem hiding this comment.
These 4 tests are all invalid, I think, based on the spec edits made in w3c/csswg-drafts#6147 for w3c/csswg-drafts#3305.
There was a problem hiding this comment.
I guess I should clarify that they're invalid, I think, because the z should be omitted in the serialization.
There was a problem hiding this comment.
This is great! But I think this should indeed be changed in a different PR which I'm happy to file after this one lands (and fix WebKit too).
It was reviewed, see WebKit bug 225019. But since WebKit doesn't yet have an auto-import process, I don't mind gathering additional review feedback. |
WebKit was a bit silly and specifically looked for a
1value in the rotation vector to determine the axis. The was addressed in r276554 with some changes tocss/css-transforms/parsing/rotate-parsing-valid.html.Of note, this new test:
This passes in Chrome and ToT WebKit but fails in Firefox which returns the value
x 400grad. I believe the test to be correct and Firefox to be incorrect.