Re: conformance (was layout solutions blah blah blah)

"Tantek Celik" <tantek@cs.stanford.edu> wrote on 1/25/02 4:34:53 PM:
>
>From: Jesse McCarthy <mccarthy36@earthlink.net>
>Subject: Re: conformance (was layout solutions blah blah blah)
>Date: Fri, Jan 25, 2002, 8:58 AM
>
>> If that's how you feel about it, perhaps you should refer to my message of
>> December 19 which I sent to this list and to you personally -- which,
>> interestingly enough, you made no response to -- that contains a URL to a
>> valid example that, thanks to the log entry included in the message,
>> _clearly_ demonstrates the statement:
>>
>> http://lists.w3.org/Archives/Public/www-style/2001Dec/0025.html
>
>Much better.  Quoting from that message (which must have gotten lost amongst
>a deluge of spam):
>
>> Did you fix the bug with @import statements including a url construct
>> containing a single quoted (maybe double quoted as well for all I know)
>> string?
>>
>> e.g.
>>
>> @import url( './modules/CSS/web.css' );
>>
>> (if the next part breaks, it is a single line from an Apache access log)
>>
>> XX.XXX.XXX.XX - - [07/Nov/2001:07:14:10 -0800] "GET /resume/val/'.
>> /modules/CSS/web.css' HTTP/1.1" 404 240 "http://www.jmmcc.
>> com/resume/val/resume.php?version=web" "Mozilla/4.0 (compatible; MSIE 5.0;
>> Mac_PowerPC)" PID4391
>>
>>
>> Or see for yourself -- http://www.jmmcc.com/resume/
>>
>> --> Jesse
>
>Now how sad is that?
>
>No, it does not appear to have been fixed in IE5.1/Mac.

>But this was simply a one line change to fix - something that would have
>been trivial to fix had we known sooner - and something which is now fixed
>in internal builds.

How unfortunate.

>In addition, the example you gave does work in IE6/Windows.

I wouldn't have expected otherwise, though I haven't had the pleasure I'm 
sure of using IE6.  It works in IE5/Windows too. 

>Thanks again for the example Jesse and please keep them coming.

No doubt.

>
>I'll use this discussion as a opportunity to plug the recently published
>Selectors Test Suite and the updated CSS1 Test Suite, both of which can be
>accessed from the CSS Test Suite home page[1].
>
>I implore folks on this list to please review these tests (in particular
>Selectors), and let the list know if you find anything missing from them
>which is present in the specification.
>
>If an example like Jesse's had been present in the original CSS1 Test Suite
>then IE5/Mac (as well as other implementations) would have better conformed
>with the feature set of CSS1.
>
>Making sure that a test suite properly provides coverage for the features of
>its respective spec is critical in helping to ensure better conformance (and
>interoperability) among implementations.
>

Perhaps a more comprehensive test suite is desirable, but it should be 
realized that it will never be definitive.  That being the case, it would be 
a mistake to use the test suite as a model for developing a browser. As I 
understand it, it is an aid to interpreting the specification and a tool for 
testing the implementation of certain discrete features, often in the 
simplest scenario.

From the perspective of implementing CSS in a browser, 
the test suite is clearly not the place to start.  You say that this 
particular bug would have been trivial to fix had you known sooner; frankly, 
I find it difficult to understand how the problem ever came into existence.  
The test suite is obviously not exhaustive, but more importantly, evaluating 
the browser with the test suite should not be considered an alternative to 
implementing the fundamental grammar of the language -- which is spelled out 
(seemingly) unambiguously in Appendix B of the spec (not to mention that the 
relevant part of this particular syntax is described in plain English in 
section 6.4). 

Typical of Microsoft the self-congratulation is voluminous and pronounced, 
but unearned.  This bug demonstrates the failure of the browser to implement 
the most basic of the basic concepts of CSS, as evidenced by its placement in 
section 1.1 of the spec, let alone represent full support; "In order for the 
style sheets to influence the presentation, the UA must be aware of their 
existence."  Perhaps paying due attention to fundamentals would have resulted 
in a better product.  To pass the "acid test" the browser has to have the 
stylesheet, doesn't it? 


Jesse

Received on Friday, 25 January 2002 21:15:04 UTC