[xsl] questions about XSLT philosophy: how much is too much?

Subject: [xsl] questions about XSLT philosophy: how much is too much?
From: "Robert P. J. Day" <rpjday@xxxxxxxxxxxxxx>
Date: Wed, 19 Mar 2003 06:53:09 -0500 (EST)
  having read two of the most recent threads -- how to find the
largest value less than a threshold, and how to find local extrema --
i'm starting to wonder whether there's a point when someone has
the right to say, "no, that's going beyond what XSLT was meant
to do."

  certainly, there are solutions to the above, but at some point,
these solutions are starting to look rather unwieldy and makes
one wonder whether the right tool is being used for the job.

  it reminds me of an ad i saw many years ago in a UNIX trade
rag -- someone was selling a personal accounting system written
entirely in shell script.  not surprisingly, the author was a 
noted shell script expert, so in his mind, scripting was the
solution to just about everything.   (when the only tool you
have is a hammer, etc. etc.)

  and the evolution of shell scripting is a good example of
creeping featurism, IMHO.  what started out as a simple 
automation utility just kept growing, as people who seemed
not to understand the proper place for a script demanded
more and more features, like arithmetic, arrays, trig functions
and so on.

  speaking from the perspective of a rank newcomer, it's getting
a bit overwhelming to see how many different problems people
are trying to solve by shoehorning them into an XSLT issue.

  as an example, consider arithmetic functions.  sure, there's
floor, ceiling, round and sum.  beyond that, if someone wants
the maximum of a set of values, i've seen the solution that
involves using a predicate to check against all values on
the preceding:: axis.

  so what's next?  adding even more math functions?  max()?
min()?  i notice in kay's book the list of exclusively 
saxon-related extensions, including more math functions,
set functions and the like.

  so i'm curious.  what is the eventual goal?  it seems
that a lot of what XSLT is being used for is starting to
push the bounds of what i initially considered stylesheet
"transformation", and that XSLT is getting strangely close
to being a fully-featured programming language.  was that
the idea in the first place?

  (and the introduction in XSLT 1.1 of <xsl:script> suggests
that a stylesheet may eventually become little more than a
wrapper for a procedural solution.)
  
  sorry for waxing philosophical.  it's early and the last
few postings just got me to thinking.

rday


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


Current Thread