Re: Type-assignment

["K.Kawaguchi" <k-kawa@bigfoot.com>]

> Well, this is a subjective issue.  You modified my algorithm so 

Right. My whole point is type assignment can be done on top of the
validator.


> Any ambiguous RELAX module can be disambiguated, I guess.  At least, 
> any tree regular language can be captured by an unambiguous tree regular 
> grammar.  (Hmm, subset construction certainly creates deterministic 

I think this may be a bit confusing.

You can "disambiguate" a grammar, only if you are allowed to alter
labels and roles. However, to alter them may change the entire label and
role of the schema, which is probably unacceptable.


> There are many ambiguity in computer science.  For example, a regular 

Ambiguity is defined in terms of interpretation. If you have two
interpretation, it's ambiguous. So you have as many ambiguities as the
definition of interpretation.


> I could not find it.  Which message are you talking about?  But we have 
> common understanding.


I think so too. But just in case you (or someone else reading this) need
it, it can be found at the following URL.

http://lists.xml.org/archives/xml-dev/200102/msg00036.html


regards,
----------------------
K.Kawaguchi
E-Mail: k-kawa@bigfoot.com