Re: TREX TC minutes from 2001-04-19

Kohsuke KAWAGUCHI wrote:

> * it assumes the grammar is unambiguous.
>   I felt that they are merely assuming the instance is unambiguous. So
>   maybe we can avoid this by making ambiguous instances invalid when
>   enforcing key/keyref constraint.

If two non-terminals compete and cannot be distinguished from the 
context, we can use their union for specifying key/keyref constraints.  
Does your algorithm (posted to xml-dev) detect such non-terminal pairs?  

Cheers,

Makoto
 
Internet: 
Nifty: VEQ00625