Re: [xsl] is XSLT 2.0 implementable? (was: N : M transformation)

Subject: Re: [xsl] is XSLT 2.0 implementable? (was: N : M transformation)
From: Tobias Reif <tobiasreif@xxxxxxxxxxxxx>
Date: Mon, 03 Feb 2003 22:48:52 +0100
Daniel Veillard wrote:


I agree that dependency on WXS is a bad aspect, but I think it won't be required for all implementations.

First news to me, how can you back-up that statement ?


http://www.w3.org/TR/xslt20/#import-schema
"
Issue 125 (schema-conformance):
We need to describe a conformance level that does not require schema support.
"


http://www.w3.org/TR/xslt20/#issue-schema-conformance
"
Issue 125: schema-conformance

Description: We need to describe a conformance level that does not require schema support.

Resolution: We decided that we should define a conformance level in which schema processing was not required. The details, however, have not been worked out.
"


I can't implement a specification I don't understand.

Then there is a lot of room for improvement regarding the spec.

  Well that would probably lead to a complete revamp of
of the structure part.


If that's what it takes to make the spec implementable (for you and probably others), then this should be evaluated by the WG IMHO.


Would Relax NG in the place of WXS solve your problems?

Of course not: Relax NG does not allow to identify type


I see; you didn't elaborate on the details so I couldn't know.


  I think Michael and Henry know me well enough, and that I propagated
that back to them. It's also clear that I tried an implementation within
libxml2 but it became quickly too painful that I focused on other targets.


My personal POV is:
I like XSLT, and I see room for improvement in XSLT 1.0 (regexen, multiple output files, etc).
So I'd be very happy to see XSLT evolve in a direction which addresses some of these areas (as the current draft of 2.0 does in some of the perhaps less controversial parts).
But all that has no value if it won't be widely implemented, which can only happen if (at leat some of the) implementers (of the currently popular processors) can implement it, and see value in doing so.



I think the spec can only improve if every implementer feeds back any problems he sees, especially if they keep him from implementing the spec.

I don't think I'm the only one :-)


Neither do I, and that's what worries me. I think the XSLT 2.0 spec might have to be changed in substantial ways. I mean, if you as maintainer of libxml/libxslt/xsltproc say

"I can't implement a specification I don't understand."

then that means that other implementers probably have rightful concerns about the current draft as well.


There is also a threshold of how much feedback seems needed to
clear up points of a spec. In most case a focused feedback makes
sense (and I know that, I have been involved in W3C working group
activities first as staff and then as external contributor since 98)
and is rightly appreciated. But the Primer can't replace an authoritative
spec where when you have a question reading it provide the answer,
that is very hard for most of XML Schemas Structure part.
Anyway I didn't want to get into this rathole, I'm just stating
why I think at this point that I won't be able to implement XSLT-2.0
nor XPath-2.0 anytime soon.


BTW, to illustrate why I enjoy some parts of XSLT 2.0:
This (unfinished, buggy, not-yet-cleaned up) transformation marks up XML code listings (eg for syntax highlighting)
http://www.pinkjuice.com/howto/vimxml/xslt/tinydbk2xhtml/markup_xml.xslt2
It's fun.



Tobi


--
http://www.pinkjuice.com/


XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list



Current Thread