RE: questionable syntax choices for XQuery

> 
> Some personal remarks about the semicolon separator between 
> declarations:
> 
> - it has been added silently in November (not listed in change list)
>   that's not nice for implementers...
> 
> - I have never understood the necessity of this change.
>   Is there some subtle lexical issue that I dont see ?

The syntax can be parsed without it, but it's very hard to produce decent
error messages and to recover from syntax errors without it. 

> 
>   actually this separator is optional in my own implementation
>   and I never noticed any parsing problem.
> 
> - I agree with Jason that making parser writing easier
>   is not a good justification for such a change.

But allowing better error messages and error recovery is a good
justification. The XQuery syntax is very fragile from this point of view
(because of the lack of redundancy), and the semicolon adds much-needed
robustness.

Michael Kay

Received on Thursday, 22 April 2004 05:16:43 UTC