Re: "CSS Standard Boasts Unprecedented Interoperability": a different - much more nuanced - opinion

Le Mer 8 juin 2011 6:30, Daniel Glazman a écrit :
> Le 07/06/11 23:49, "Gérard Talbot" a écrit :
>


>> The fact that 9000 CSS tests have been submitted and that the fact the
>> CSS
>> 2.1 spec is official do not, by itself, make CSS coding entirely
>> reliable,
>> perfectly predictable in all browsers.
>
> And we never said CSS is entirely reliable or perfect. You perfectly
> know the Test Suite is here to test the features in a spec written by
> humans. To the best of our knowledge, the CSS 2.1 Test Suite tests
> all features in the spec.


Daniel, please visit this page:

CSS 2.1 test suite: Known issues
http://wiki.csswg.org/test/css2.1/issues

and I assure you that such page is not completely updated right now.

[Addendum: latest update: in my opinion, vertical-align-117 and
vertical-align-118 are not really and not accurately testing what those
testcases were originally aiming at testing. So, those 2 testcases should
be upgraded, improved.]



> It is just impossible to test *all*
> interactions between *all* properties and *all* values in *all*
> scenarios. The errata mode is here to cover the issues raised after REC
> release.
>
> </Daniel>
> --
> W3C CSS WG, Co-chair


If you are unaware of a failure or unaware about a bad testcase or unaware
about some interaction between a property and a value and a webpage
scenario, then there is little you can do. I am not asking anyone to be
omniscient.

What about when a failure, a bad testcase or a precise interation or
scenario is known, has been identified publicly, has been documented, has
been explained, in some cases, has been discussed, etc? This is what I was
talking about and why I mentioned

http://wiki.csswg.org/test/css2.1/issues

2 examples.

1-
On october 18th 2010, David Baron wrote

http://lists.w3.org/Archives/Public/public-css-testsuite/2010Oct/0280.html

saying that 12 testcases are invalid. Today, those same testcases (RC6 and
in nightly-unstable) still have not been corrected accordingly.

2-
On 7 Nov 2010, I explained in
http://lists.w3.org/Archives/Public/public-css-testsuite/2010Nov/0039.html
that user agents can render a missing icon if the targeted image is not
fetchable. Therefore, either the testcase
http://test.csswg.org/suites/css2.1/20101027/html4/before-after-images-001.htm
needed to be upgraded or corrected to reflect such reality in the spec and
in some browsers. Today, that testcase still has not been corrected
accordingly.

I mentioned 50 invalid testcases: they all have been identified,
explained. Fair and square.

I mentioned 200 testcases requiring some code tuning, tweaking,
corrections, improvements: they all have been identified. Sometimes as
soon as fall 2010.

Etc.

Gérard Talbot
-- 
CSS 2.1 Test suite RC6, March 23rd 2011
http://test.csswg.org/suites/css2.1/20110323/html4/toc.html

Contributions to CSS 2.1 test suite
http://www.gtalbot.org/BrowserBugsSection/css21testsuite/

Web authors' contributions to CSS 2.1 test suite
http://www.gtalbot.org/BrowserBugsSection/css21testsuite/web-authors-contributions-css21-testsuite.html

Received on Monday, 4 July 2011 20:05:07 UTC