Lists Home |
Date Index |
I don't see how one can consider a language design that doesn't
allow extensions if it is to be a living language. Conformance
has to be profiled in most cases anyway to enable implementations
of subsets. Are you considering it to be a lockstepped monolith?
If we didn't enable slang, Americans would speak English. That
would be a disaster for the west coast media.
From: Jonathan Robie [mailto:email@example.com]
I agree with this. We are just starting to explore conformance issues
seriously in the WG, and I do believe that the goal of interoperable
implementations is essential.
And sometimes tricky. For instance, are vendors allowed to do extensions?
It's easy to say no, but what if some vendors want to implement updates
before XQuery has added them to the language - are they really not allowed
to add updates as an extension? If they *are* allowed to, should we require
that there be an option that complains about every vendor extension?