Re: My favourite XSLT enhancement requests

> There are many proposals for additional features that would convert XSLT
> to the complete (and let's be honest, the *procedural* rather than
> *declarative*) programming language.

Personally I am not interested in that great a revision.  I think the
language as it is now is very good.  I'd prefer some syntactic sugar and
added power in a few wisely chosen places: something like xsl:function
for another way of having a named template -- not even escaping to a
scripting language, and perhaps evaluate(). (And the already agreed
conversion of RTF to node-set and known issues like regexps.)

I think dynamically scoped parameters would be useful, but I recognise
they have drawbacks.  Someone wiser decide if they should go in.

If anybody knows how to get back to the original documents once you've
done apply-templates on a document(), I'd appreciate a hint.  Best I've
been able to come up with is two passes, using RTF->node set to gather
all the necessary interesting details into one big variable and consult
that as I go.  Then again, that might give the best performance anyway
since I'll have to generate index, glossaries, biblio and TOCs on the
fly anyway. :-/  Multi-document() keys/indices with ability to restrict
them to node subsets on lookup would be a help.

Hmm, this is starting to sound like I really ought to be using a
database and one of the database-aware XSL engines...  But for now, I
need to stick to files.

Cheers,
//lat
-- 
Whoever said that love is blind was dead wrong.  Love is
the only thing on earth that lets us see each other with
the remotest accuracy.  --Martha Beck


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