RE: [xsl] xslt and fuzzy logic

["Bryan Rasmussen" <bry@xxxxxxxxxx>]

>Well, XSLT 1.0 was not designed with heavy computation as a primary use
>case.
I know, I was thinking about the computational model of xslt 2.o, I happen
to like the idea of doing at least some computation in xslt as it is a
functional language. I like functional languages for heavier computation


>Nor was XML designed to provide a syntax for "languages" in the sense of C,
>Java or Perl -- or XSLT -- but rather for *markup* languages, which is
>quite a different kind of critter. If you have an application that can
>really make use of semi-structured data and can use XML for storing and
>managing it, this balance may be easier to find. There's a big difference
>(even given an interesting overlap) between designing the structure of a
>data set, and designing a grammar to be interpreted or compiled for
execution.

>The overlap is in using XML to design a processing specification or even,
>like XSLT, a declarative language. I don't have experience in this area
>(I'd like to hear from readers who do) -- but XSLT itself "approaches the
>nature of a scripting language" doesn't it? (And there are those who feel
>its XML syntax is, on balance, a liability not an asset. I say "well it
>depends" -- again, on the problem you're trying to solve and how close that
>is to the class of problems XSLT was designed for.)

well yeah it seems to me that a lot of the specialized little languages that
people used to write for applications and so forth seem to be written in xml
now.

yes XSLT does approach the nature of a scripting language, that's the
difficulty I was implying, it's hard to design a language that creates an
instance of the language with which you're interpreting the designed
language! There are some that are useful, schematron for example, it's even
harder I would surmise, to do one that does the same task that the
interpreting language is designed to do. You see this in the examples of
xslt books all the time, use xslt to output xslt! the input xml documents
are often too simple to be useful, if you try to make an input that is more
complex then you start to design a language, with the nature of a scripting
language, that transforms xml. in other words you make xslt. For example I
made one for some of the content writers, designers at our place, so that
they could generate most of the templates for a new project, the girl using
it most of the time, ended up by learning how to write the xslt it output
anyway, by sitting beside me for a couple of days.
there have been some that are useful, for example the loopxsl compiler,
where people could write iterative loops and it outputs recursive loops for
them. a time saver if you want to have somebody used to iterative
programming write some parts of your xslt for you.
All this said I do a lot of generation of stylesheets from meta-documents, I
bet a lot of us do the same, but such meta-documents are often project
specific.



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