Lists Home |
Date Index |
>The trouble is I'm not sure that the experiment would provide any useful
It seems that you're thinking of a rather different kind of
experimentation from me. I was thinking it would answer questions
- does it make it easy enough to author (in the way that adding
a reference to an external parameter entity apparently doesn't)?
- is it extensible by the user, or fixed for a given vocabulary?
- does it deal with the conflicting uses of the same name by different
vocabularies (eg by using qualified entity names, or making entities
visible only in certain scopes)?
- is it compatible with existing software (eg not requiring a new
kind of XPath node)?
- does it satisfy the claimed need for characters that aren't in unicode?
- can it be used independent of any particular schema mechanism?
- does it cause enormous parser bloat?
- can it be effectively used in browsers (eg can they cache the
entity definitions for HTML)?
- is it round-trippable (can serializers reasonably put it back)?
It seems to me that different combinations of the issues above are
important to different groups, and we need concrete implementations so
they can assess whether the compromises needed still produce something
worth the change.