> Funny enough, these three SAXON functions *do not*, at my
> opinion, break the XSLT processing model in any substantial way. [...]

I think saxon:assign definitely breaks the model in some sense, or at least
reduces its purity. Of course, there are functional languages that permit
assignment, and I think one can define semantics for saxon:assign that are
perfectly well-defined; but it's certainly true that the presence of
saxon:assign imposes a sequential ordering that wouldn't otherwise be
required.

I can't see any problem with saxon:return, provided one is careful to define
unambiguous behaviour when multiple saxon:return statements are
instantiated. I thought I had designed the rules so this wasn't allowed, but
Jeni has shown I didn't get this right.

I'll leave the formal semantics to someone else: I'm an engineer, not a
mathematician.

Mike Kay
Software AG


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list