Re: [xml-dev] Schema style and parser performance

[Richard Tobin <richard@cogsci.ed.ac.uk>]

>If validating and converting the schema to internal form is a large part of
>validation cost, would transformations that "flatten" a schema make a
>significant difference?
...
>I am thinking of things like
>flattening type derivations, putting the contents of included schemas
>inline, etc.

Certainly type derivation is one of the most complicated parts of
schema compilation, but I don't know whether it's expensive in CPU
time.  XSV is full of inefficiencies, and when we've profiled it it
has usually turned out that lots of the time is spent in some piece of
code written without any thought for efficiency, or some idiom that
turns out to be expensive in Python.  I've never profiled a schema
validator written with efficiency as an important goal.

-- Richard