RE: [xml-dev] XML support in browsers?

["Michael Kay" <mike@saxonica.com>]

> Microsoft made a perhaps well-intentioned mistake of 
> implementing a beta version of the XSLT 1.0 spec and shipping 
> it in IE 4. 

Yes, that was a bad mistake.

> They then made then horrendous decisions to 
> continue supporting the incompatible beta version for years,

No, I don't think that was a horrendous decision. Given the earlier mistake,
I think people who write web content that relies on a general release
Microsoft language have a right to expect a decent period of notice before
their content stops working.
 
> and to stop all development on conformant versions. 

They didn't stop development of MSXML until they had shipped a complete and
largely conformant implementation. (The complaints about non-conformance
relate almost entirely to areas outside the scope of the XSLT specification,
such as whitespace handling in the XML parser, mapping of the XPath data
model to DOM, and interpretation of MIME types.)
 
> They have 
> continued to prioritize conformance to old, buggy versions of 
> Microsoft products over conformance to the actual specifications.

No, they haven't. They stopped development of the old WD-xsl as soon as XSLT
1.0 became a Rec, and did their best to let the world know it was
obsolescent. In the 2nd edition of my book, April 2001, I reported that it
was becoming very difficult to find the documentation for WD-xsl, because
Microsoft was promoting XSLT 1.0 in its place. MSXML3 shipped with the
WD-xsl processor but no language specification.
> 
> That's rewriting history. Actual XSLT 1.0 support arrived in 
> other browsers well before anything approximating it showed 
> up in IE. (I'd argue it still isn't there in IE 8.)

Sorry, but that's wrong. In April 2001, I wrote that MSXML3 (with XSLT 1.0
support) "was first released in March 2000 [that's 4 months after the spec!]
and then followed by a rapid sequence of releases until the first production
release in October 2000", and that it was expected to be shipped with IE6.
In the event the general release of IE6 was in August 2001: that's less than
2 years after the spec, which seemed a long wait at the time, but is pretty
fast by today's standards. And although I was very pernickety in the book
about non-conformances, I found very little to criticise them on.

As far as I can determine, the second browser to support XSLT 1.0, other
than through user-installed plug-ins, was Netscape 7.1, released in June
2003. And if memory serves me right, it took a long time before it reached
comparable levels of conformance. 
> 
> White space handling is a continuing problem, and hardly minor.
> Incorrect, non-standard MIME types are another. These should 
> be easy to fix, but Microsoft's persistent refusal to fix 
> them 10 years in makes me seriously question bother their 
> competence and their commitment.

Both are serious, but neither is an XSLT conformance issue, and neither, I
believe, has significantly affected the timeline. And in both cases,
(especially the MIME types), fixing the problem would cause huge disruption
since many people have built systems or created content that depends on the
bug being present. This is one of those areas where the web is so fragile
that it's difficult to do anything without causing more damage than you
reckoned on.

I'm not usually a fan of Microsoft. But in this instance, I think you've got
it wrong.

Regards,

Michael Kay
http://www.saxonica.com/
http://twitter.com/michaelhkay