Best practice in validating db5 with relax-ng?

["Francesco Pretto" <ceztkoml@gmail.com>]

Hi, I'm a newcomer, greetings to all! I'm just switching from Docbook
4.4 and I was wondering a couple of things:

1) how I should validate a large document splitted in several files
using MSV tools (relames.jar), as specified in the howto? Should I
validate all the xml files separately or is there a way to do
validating while processing XInclude directives (sample commands are
largely appreciated)?
2) In general, should I perform always validating before any
processing (XInclude and/or xslt)?

I've even an unrelated, probably OT, question: I appreciate the choice
to move to relax-ng, but will ever come the time relax-ng schemas will
be highly integrated with libxml2 tools like DTD schemas were? For
example, I very liked the concept of "installing" schemas in
/etx/xml/catalog . I'm pretty sure the success of Docbook+relax-ng is
highly related to their integration with common tools you can find in
any linux/*nix installation (xsltproc, xmllint, ...).

Thanks for any suggestion.

Greetings,
Francesco