Re: Issue: key/keyRef and modularity

[James Clark <jjc@jclark.com>]

Murata Makoto wrote:
> 
> Norman Walsh wrote:
> 
> > For 1.0, I'm hesitant about adding the complexity of renaming or
> > scoping of keys. I think (although I admit I've only thought about it
> > for 30 seconds or so) that I would prefer to note this as an issue and
> > leave it unresolved for RELAX NG 1.0.
> 
> Seconded.  In general, if we have independently developped grammars or
> patterns, include will cause conflicts.

I'm also hesitant about adding any complexity.  However, I'm still
concerned. The *only* thing that causes conflicts when using externalRef
to include independently developped patterns or grammars is key/keyRef. 
I believe closure is an important feature of RELAX NG.

I wonder whether it would help if we specified that the symbol space of
the key or keyRef is identified both by the value of the key/keyRef
attribute and the inherited value of the namespace attribute.

James