Microsoft, you should be ashamed
By glazou on Thursday 18 March 2010, 15:01 - Microsoftisms - Permalink
Yes. And despite the very nice IE 9 preview that shows real progress on Web Standards. I just found that page that shows, holly miracle, that MSIE9 Preview passes 100% of the tests written by... the IE team and submitted to the W3C. Isn't that cool? Isn't that cool to show that IE9 passes these tests while other browsers don't? It easily forgets that there are thousands of other tests that IE does not pass, that there are tons of specs that Mozilla, Webkit and Opera all implement but not Microsoft...
Not only that but the tables and figures given for non-MSFT browsers are often totally wrong. I just ran the Selectors (MSFT ones) tests with Safari 4.0.5 and all the tests are green. In the summary table, Microsoft says Safari 4.0 passes only 44%... I also ran the DOM Level 2 Core tests with Firefox 3.6 and it shows 100% green while MSFT says 89%. Some tests even say Firefox does not support 'border-radius' but the property is STILL vendor-prefixed at this time so the test is irrelevant! With the vendor-prefixed version - and that's the only one existing at this time - Firefox would show a 100% support...
Even worse, some tests, like the "@import inside @media" test, seem to me wrong. '@media screen { @import "foo.css" }' should result in one @media at-rule with no valid style rule inside. It's then perfectly normal to have document.styleSheets[0].cssRules.length == 1...
Or test 'syntax for background-repeat' is also wrong because getComputedStyle() is called on testDiv but testDiv does not exist in the JavaScript context !! The 'var testDiv = document.getElementById("testDiv");' line is missing !!!
Let's take now test ':nth-child() and comments'. It says wrong for FF but the selector tested is
div div:nth-child(/*COMMENT*/2/*COMMENT*/n/*COMMENT*/+/*COMMENT*/1/*COMMENT*/)/*COMMENT*/
and that is wrong because CSS grammar says comments are valid anywhere BETWEEN TOKENS and the CSS WG has always discussed the "2n" here being tokenized as a dimension. We never ever thought of an author being able to insert a comment between the number and the "n". It probably does not make sense anyway and it's probably something the CSS WG should clarify because allowing comments here will again complexify nth-child() parsing that is already complex enough, and MSFT should not test at this time before clarification - and at least decision - happens.
Same thing for test "A negative border width property should not be valid" !!! Did they check the JS console performing the tests ?!? How many wrong tests in their list !
WTF?!? Who dared writing a browser chart like that page? Jason Upton, who signed that page, I don't know you but I don't like at all your methods, they stink like the darkest moments of 1997/1998 at Microsoft.
Come on Microsoft, clean a bit that page or suffer other tests comparisons...
Update 19-mar-2010: SVG in IE9...
Comments
And given the fact that all other browsers are released versions (and not a "platform preview" as is IE9), a little bit of honesty would have been to also include the current release of IE (ie. IE8) to show how it fares against current browsers.
With the wonderful Internet Explorer, you can use "testDiv" straight away if there is an element with this id exist on your page. There is no need to call document.getElementById() in your JavaScript.
It's not a bug, it's a feature for lazy web developers...
It looks again like this "Get the facts" campaign, this crappy business way of doing things. Completely dishonest and disgraceful, but we start to be used to it now... unfortunately. Sniff.
@djano: absolutely ; so the Web Standards-compliance tests are not themselves Web Standards-compliant. I hope everyone appreciates the beauty of the thing :-)
"Some tests even say Firefox does not support 'border-radius' but the property is STILL vendor-prefixed at this time so the test is irrelevant! With the vendor-prefixed version - and that's the only one existing at this time - Firefox would show a 100% support..."
That does mean Fx does not support border-radius, so that's not really wrong, is it. If I were making a table like that I'd probably stick a pale green with an asterisk and at the bottom "*vendor prefixed" or some such if a particular test passed with the vendor prefix.
Btw, I don't know if you've seen http://my.opera.com/haavard/blog/20... but it's certainly quite revealing to see the real SVG support at http://www.codedread.com/svg-suppor...
heheh I think the embarrassment should be the owner of microsoft
I don't know how you performed the tests but with Safari 4 and Firefox 3.6 I got the same results as Microsoft's...
gregseth, some of the tests are _wrong_, which means the results are invalid. They're displaying an MS DOM bug rather than whether or not other browsers support certain CSS properties.
One more thing. I checked the latest Chromium build with the "@import inside @media" test and it appears that they FIXED THE BUG (i.e. it now also returns one object CSSMediaRule as it should). So according to Microsoft's stupid test, Chrome is now less compliant than it used to be. Good God.
Check out the IE blog, they linked to your page and said they have updated testcases.
http://blogs.msdn.com/ie/archive/20...