[relaxng-user] Newbie: Normalisation and PSVI

[Tor Helland]

> > We may define element values as string or token, and more if using
datatype
> > libraries. And when using token for elements, the element values will be
> > normalised.
>
> There is no point, though it is valid, in using either "string" or "token"
> as patterns by themselves, since neither one constrains the input.
> These datatypes are useful only in connection with a specific value.

So a token datatype does not mean anything except in enumerations, with
<value> definitions, since the instance text and the <value> text are both
normalised before comparison.

I had misunderstood that, putting relax ng in the role of managing the
whitespace normalisation process.

Making a utility that actually did that, using the relax ng schema to
normalise whitespace and streamline other datatype text representations,
could be really useful. Does any such utility exist? (But to exploit that
utility a schema would probably be built with much datatype info that would
be pointless during the normal relax ng validation.)

-tor