RE: [xsl] A beef with XSLT Sometimes too complicated

[Wendell Piez <wapiez@xxxxxxxxxxxxxxxx>]

At 11:45 AM 7/18/2006, Mike wrote:
> > It does occur to me that what we are facing is a collective
> > coming-to-grips with the question of
> > queries-as-transformations and transformations-as-queries,
>
> Queries extract a small part of the information available from some
> resource. Transformations convert the representation of information into a
> different form. It's quite possible, therefore, for a task to involve both
> query and transformation.

Indeed. An extraction of a list of phone numbers might be a "pure" 
query. A plain-vanilla reformatting of a Docbook page into HTML might 
be a "pure" transformation, or close to it. But most tasks involve 
something of both.  Right in the middle might be the creation of a 
table of contents into a set of XML documents as rendered in HTML.

Not only that, but to addle our brains further ... sometimes it's 
rather difficult to distinguish between the information and its 
representation. The representation of information is itself 
information (which is why a well-designed, scalable information 
processing architecture will probably introduce some kind of 
layering). In that respect, a transformation is a translation of one 
kind of information (say, Docbook tagging) into another (say, the 
code that drives a renderer to display the same "information" as the 
source markup in a form friendlier to the eye and the educated 
brain). But translation is rarely a simple mapping; it usually 
involves the removal of some information (say, the nesting becomes 
implicit in the level of headers rather than explicit in the markup 
of sections) while other information is added (relative font sizing 
is introduced for the display of those headers) -- generally, one 
hopes, according to consistent rules derived from the relevant 
context, which (in principle, if not too much is thrown away) may 
even enable the translation back again.

This may seem like idle philosophizing until one considers that this 
is effectively what queries do too, only balanced somewhat 
differently with respect to adding and throwing away.

A query is a transformation scoped to a "repository" and returning an 
extract of it. A transformation is a query scoped to a "document" and 
returning an elaboration of it.

Cheers,
Wendell