Re: [xsl] New XSLT 3.0 Working Draft

[Michael Kay <mike@xxxxxxxxxxxx>]

On 10/07/2012 19:01, Michele R Combs wrote:
> Since relatively few folks adopted 2.0, what do we think the chances are of 3.0 being adopted?
It's an excellent question. But I think it needs to be expressed in 
terms of "when", not "whether". XSLT 2.0 appeared in 2007 and there has 
been roughly one new processor appearing every year. There are still 
environments on which no XSLT 2.0 processor is available, but they are 
becoming less common. The most recent development is Saxon-CE which 
brings XSLT 2.0 to the browser for the first time.

Of course cost is a factor. One reason people stick with XSLT 1.0 is 
that nearly all the processors were free. Very few people have been able 
to justify writing an XSLT 2.0 processor and giving it away for nothing; 
most of the XSLT 2.0 processors therefore cost money, and however much 
better 2.0 is, there will be a lot of people who find it hard to 
persuade their boss to allocate budget. Equally, the main environment on 
which XSLT 2.0 is still not available is the LAMP platform, and the main 
reason for that is that users of this platform are markedly reluctant to 
spend money on infrastructure software; without funding, there will be 
no investment.

Certainly I think you can expect a long wait before XSLT 3.0 is widely 
implemented. To me it feels a little bit like the SQL scene, where it 
took around 10 years from publication of SQL 1992 to widespread 
availability and adoption (and even then, the actual level of 
conformance was variable). I'm not sure how W3C processes are going to 
adapt to this, because they still seem geared to the old idea of 
"internet speed" in which you don't regard a standard as finished until 
the implementations are available; what is clear from the XSLT 2.0 
experience is that most implementors won't start work until the standard 
is stable.


Michael Kay
Saxonica