xml-dev - Re: Procedural vs Declarative XML transformation approaches

[Rick JELLIFFE <ricko@geotempo.com>]

Rick JELLIFFE wrote:
> 
> > Mike.Champion@SoftwareAG-USA.com wrote:
> 
> >  I can think of a couple of arguments.
> > First, In the world of query languages, it's clear to me that
> > procedural approaches don't scale well -- you want an optimizer to
> > figure out the best way to "just do it" given all it knows about the
> > data, indexes, storage structures, etc.
> 
> I don't think that it has been established that what might be clear for
> relational data (tables) is also true for XML (trees).
> 
> I  can see that if one were accessing tree data as if it were a
> hierarchy of records and one had to explicitly nominate the records to
> be taken (i.e., if each subtree had to be grown or culled) then some
> declarative model would be useful.  But I don't think that is how XML is
> used: One starts with the fact of a tree in which the most interested
> data is already contained in nice branches (rather like pre-indexing?);
> if one starts with a graph-of-records model then there is no natural
> boundary.
> 
> In other words, for relational data, data is grouped either by being in
> a records or by keys.  In XML data, it is groups by elements, by path
> expressions or by keys.  That additional second stage (being able to
> express relations by constant paths along a tree) is declarative, and I
> think it changes the boundaries for querying to an extent that saying
> "declarative scales; procedural doesn't"  because even procedural
> queries will have significant declarative components.
> 
> Rick Jelliffe