search

LeaVerou / css3test

How does your browser score for its CSS3 support?
http://css3test.com
MIT License
214 stars 83 forks source link

Add all predefined counter styles #207

Closed Zefling closed 2 years ago

Zefling commented 3 years ago

Problem with Firefox because of @counter-style support I can't find solution.

I'm not sure if the tests are really pertinent.

SebastianZ commented 3 years ago

Problem with Firefox because of @counter-style support I can't find solution.

You mean because Firefox recognizes all values of list-style-type as valid? I don't know either but this could be added as a note under "Cheaters" in the sidebar (even when it's actually not cheating because any identifier is valid at parse time).

Sebastian

Zefling commented 3 years ago

I wanted to try to add a test with the retrieve of the LI value , but nothing in the DOM seems to allow it.

SebastianZ commented 2 years ago

@PolariTOON @LeaVerou What do you think? Should they be removed? I'd say they could be added back if we added a way to make them optional, i.e. not let them count towards the overall percentage.

Friendly ping on this question.

Sebastian

PolariTOON commented 2 years ago

Sorry for not replying earlier, i thought i did it already.

Counter Styles Level 3 says:

If a <counter-style-name> is used that does not refer to any existing counter style, it must act identically to the decimal counter style (but does not compute to decimal).

Both Firefox and Chrome seem to implement this behavior while not supporting most (perhaps any?) value from Ready-made Counter Styles. I guess that's why all the tests pass in these browsers. Webkit does not seem to accept every identifiers, but just the ones it really supports. This includes some of the values defined in this note, but unknown values are just treated as invalid and thus do not generate a decimal counter. That's why we are able to detect support in Webkit, but it's actually a non-standard behavior.

Given that we should not be able to test the support for these counters in a conformant browser, that this note is non-normative, that apparently we don't use to test optional features (such as invert for outline-color or aural properties defined in CSS2), i also think we should not add this note to the test suite.

Zefling commented 2 years ago

Same opinion as @PolariTOON , if all the tests are faked on Blink and Gecko, it only gives an incorrect result. As long as there is no real solution to differentiate it, I don't think it's worth for merging.

SebastianZ commented 2 years ago

Ok, so I close this PR then.

Sebastian