Re: Issue: TREX elements appearing in foreign elements (Re: ATREX<documentation> element)

[Murata Makoto <mura034@attglobal.net>]

>I've actually implemented TREX using a SAX interface.  In my
>implementation, allowing TREX elements inside annotations does not
>require extra code; in fact disallowing TREX elements inside annoations
>would require extra code in my implementation. This is objective fact,
>not subjective opinion.
>
>Parsing TREX patterns requires you to deal with elements
>context-sensitively; elements are handled differently
>
>- within a grammar element
>- when expecting a pattern
>- when expecting a name class
>- within an anonymous datatype
>- when handling an empty element
>
>Whatever mechanisms the program uses for dealing with this
>context-sensitivity can also easily deal with ignoring TREX elements
>inside annotations.

In other words, any program that handles TREX patterns is required to 
deal with elements context-sensitively, even if the program merely 
replaces <p> with <para> (or <danraku> in Japanese).  Can a simple 
XSLT stylesheet implement such renaming? 

>This is totally backwards.   There is a distinction between
>
>(a) Complexity of the language as it appears to a user from the
>specification
>(b) Complexity of the implementation

You tend to think that no restrictions are easier to understand for users.  
On the other hand, I tend to think that tight restrictions are easier to 
understand for users.

Cheers,

Makoto