OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   RE: [xml-dev] Re: XML and Complex Systems (was Re: [xml-dev] Re: An Arc

[ Lists Home | Date Index | Thread Index ]

>[Eliotte Rusty Harold]
>Much more important is that his test case is:
>XML documents contain (among other things) text phrases that must be 
>converted into equivalent LaTeX phrases. Some text phrases, such as "&" 
>and "$" have special meaning to LaTeX and thus must be escaped during 
>processing. Others represent text idioms like "(C)" that must be mapped to 
>their LaTeX equivalents ("\copyright{}").
>In other words he wants to do string manipulations on unmarked up text. 
>Furthermore, his output format is not XML, but LaTeX. Moertl is taking 
>XSLT and using it to do exactly what it was designed not to do. He is 
>completely confused about what the intended purpose of XSLT actually is. 
>It was never intended to do what he wants it to do. It shouldn't be a 
>surprise he has trouble. Nor should this be considered a knock on XSLT, 
>since none of his use cases are something XSLT was ever intended to handle.

Oh I dunno. I believe he has a very good point. XSLT has trouble 
manipulating things that are between the tags.
In any evolved (read "succesful") markup vocabulary most of the interesting 
stuff is *between* the tags. This
is basically Herman Zipf's "Principle of Least Effort played" out in markup 
languages. I have written
about this in this months XML Journal ("Soft Issues surrounding Industry 
Standard Schemas" http://www.sys-con.com/xml/)

Any XML transformation system that limits itself to explicit structural 
is of limited use in the real world.

There are those who would argue (and I was one of them in a previous life) 
that the way to fix this
is to provide an "escape" into a procedural environment - be it an embedded 
scripting language
or an escape to roll-your-own extension functions.

I no longer believe this is a good answer and, as I said at Paul Prescods 
excellent XSLT
talk in Orlando, I believe the answer lines in pipeline architectures which 
the mixing and matching of different XML processing paradigms in a single
execution context.

BTW, I have done some analysis which shows a good fit between the frequency 
of element types (tags)
and Zipf's inverse square law http://www.wikipedia.com/wiki/Zipfs_law in 
XML corpora.
I'm searching for other work in this area - any pointers appreciated.



News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS