Re: Name classes

[Murata Makoto <mura034@attglobal.net>]

Kohsuke KAWAGUCHI wrote:

> I think you should send it to TREX ML, Murata-san.

Since you didn't, I didn't.  Your message was sent to me only.  

> I know that every automaton can be written back into regular expression
> *theoretically*. But there are just so much sugar in XML Schema that
> makes this transformation impractical.

I don't understand.  I was not talking about XML Schema.

>And if the resulting regexp has
> to be somewhat legible otherwise the schema itself becomes meaningless,
> doesn't it.

Yes.  But I was not talking about convering the result of regexps back 
to schemas in my first mail.

> > If we consider validation only, there are no problems.  
> > 
> > If we consider operations to schema, we already have many problems.  Why 
> > not another?
> 
> I think this one is an unnecessary complication because I don't think we
> need regexp to restrict tag names.

My original post was a request for criteria.  Certainly, some regular exprssions 
for tag names are useful.  E.g.,"any attribute names up to 10 characters" 
and "any attribute names in US-ascii".  Do they hit 80%?  Doubtful.  But we have 
to have better understading of the requirement.

Cheers,

Makoto