Re: [DM] BEA_006

> (XSLT could very easily enforce that
> documents nodes have to obey Infoset rules).

Only by rewriting history, and removing one of the design requirements
that went into the design of the language, and a functionality that has
been used by many users since the language was released.

> With this respect we are in a dilemma: we are either
> backwards compatibility with XSLT 1.0 or in compatibility
> with XML itself and Infoset.

Not at all. The grammar and definition of an external parsed entity is
defined in the XML REC and is (unlike the infoset) a core feature of the
XML recommendation.


The infoset chose only to represent complete documents, That is their
choice but that came some years after XSLT. The fact that they made that
choice (which may well be reasonable in the context of the infoset) can
have no bearing on the model used for Xpath. XSLT was designed to be
able to generate external entities and that's what it is used for.

Already with Xpath 2, W3C is in danger of losing all credability as a
standards body as backwards compatibility has clearly been given such
little importance. As far as I have seen (after several threads on
xsl-list) xpath2 is already so unpopular amongst existing Xpath/xslt
implementors that the majority are not planning on implementing xslt2 at
all. At this stage in the process for XSLT1 there were several available
implementations that could be tested and allow comments on the language
features. Don't you find it worrying that currently there is only one
XSLT 2 implementation publicly advertised?

The working group should be striving to remove many of the more glaring
incompatibilities and usability problems, not considering adding more.

David

Received on Wednesday, 18 February 2004 15:51:35 UTC