Now the top four browsers have either released or announced a version of their browser that passes Acid2 work on Acid3 has restarted and Ian Hickson has been adding testcases over the last few weeks. Acid3 is primarily concerned with ECMAScript and the DOM though Selectors Level 3, Media Queries, and data: URIs are also tested. The idea of the test is to run a hundred ECMAScript functions that either return true (pass) or false (fail). I believe sixteen of those functions are missing at the moment and people are encouraged to contribute. Tests need to be justified by a specification at W3C Candidate Recommendation, W3C Recommendation, or equivalent for non-W3C consortia, and be from 2004 or before. I believe that if browsers once again commit to fixing the bugs found by this test Web authoring will improve tremendously.
You're also missing one further requirement: at least one of Firefox, Opera, or Safari must fail the test.
I would have really liked a test for HTML4 presentational attributes of <col>, especially align (Mozilla bug 915) to be included in ACID3. But it is difficult to write rendering tests for anything but the simplest cases as standard ECMAScript functions. I only know how to test for computed styles when using just the standards. Standardizing MSIE DOM extensions such as clientTop, offsetWidth, etc. and getting them supported across browsers would help make that possible. I wonder if anyone can come up with an alternative way to do it.
@Gaurav: The intention is to allow the HTML 5 test-suite to cover HTML issues such as that, and the CSS 2.1/3 test-suite to cover CSS).
Any consideration of covering XHR? It seems to fall within the parameters you describe.
Sorry, I should think before I post; I thought it would be applicable because HTTP falls into the criteria, but of course XHR itself doesn't.
I believe bits of HTTP are tested through other means, but this is indeed limited due to not being able to use XMLHttpRequest.
Bring on Acid 3!
For Acid 4, please consider including Scalable Vector Graphics (SVG).
I'm curious about where to find the specification to support test 91 (property enumeration order).
Please include some MathML (became recommendation in 2003) in the test.
Oops, never mind. Test 91 (now?) doesn't actually rely on some implicitly defined ordering of properties, as I previously thought. All is well with the world.
You're also missing one further requirement: at least one of Firefox, Opera, or Safari must fail the test.
Necessarily? Isn't this also intended as a tool to get Internet Explorer to catch up?
What about trying to verify the good behaviour of <button value=""></> ?
MSIE7 is still the worst to do that.
Oh. And also : XmlHTTPRequest with MSIE 6 and 7 , in GET method send the wrong charset, even if UTF-8 is asked. POST method works well.
When it comes to your list of allowed XHTML elements, it should be KBD, not KDB.
@Da: XMLHttpRequest is out of question — it isn't even now a CR, yet alone in 2004.
It would be awesome if it could also test nested css counters with different list-style-types, to push browser vendors to support them better!
Web fonts should be tested. Please!
@fonty: If I am reading the latest Acid3 work-in-progress source correctly, you get your wish. It seems to require browsers to be able to download and use the font Ahem. Although I would have thought this would be out of scope for Acid3 since CSS3 Web Fonts is only in the Last Call stage, and CSS2 had been superseded by CSS2.1 which removed web fonts. Of course, the reference rendering hasn't changed though, so maybe it doesn't and I just haven't figured out why.
People, people... Be careful what you wish for! Fully supporting HTML may create far more trouble than you think!
For example, did you know that it's only because HTML isn't implemented like it should that you can actually use XHTML? Remember, short-tag syntax ( <tag/content/ ) isn't enforced by any browser - and thanks to that, you can use stuff like <br /> to display XHTML as HTML with no trouble.
The Acid tests are a good way to ensure that various implementations of currently supported major elements (and error handling) do match one another reasonably well (and preferably match existing, ratified standards), they don't directly cover actual standards support.
whoa, it actually crashes opera 9.26
WebKit passed Acid3 fully for the first time on Friday. So it's 100 subtests of 100 passed now.
Opera 9.60 Beta scored 85.