Hi,

<musings> 
If compression is characterised as "encoding information while reducing the
bandwidth or bits required" [1],
and note that in lossy compression "The lost information is usually removed
because it is subjectively
less important to the quality of the data" [2].

Then isn't it possible to characterise a transformation of TREX (or RELAX or
XML Schemas) into
something terse like Hook [3] as lossy schema compression?

Hook is 'ultra-terse' and so is 'compressed' in the general sense. However
it is also
lossy because it is coarse-grained.

The loss in this case is in the available 'semantics' (and I'm using that
term recklessly) of the schema validation.
However in some (many?) cases the essential features of the desired
constraints
may be retained.
</musings> 

A different perspective, or has this been covered elsewhere?

[1]. http://www.dictionary.com/cgi-bin/dict.pl?term=compression
[2]. http://www.dictionary.com/cgi-bin/dict.pl?term=lossy
[3]. http://www.ascc.net/xml/hook/

Cheers,

L.


--
Leigh Dodds, Systems Architect       | "Pluralitas non est ponenda
http://weblogs.userland.com/eclectic |    sine necessitate"
http://www.xml.com/pub/xmldeviant    |     -- William of Ockham