On Oct 17, 2006, at 12:46 PM, Subbu Allamaraju wrote:
The key point I would like to make is that XHR is more abstract
than what you suggest, and there are use cases that can be solved
by creating APIs layered over XHR. In those cases, the layers
should be able to define method support applicable at that layer.
Secondly, it does not make sense to lump all possible
implementations into one class and require all those to be inter-
operable.
If your implementation won't be interoperable, then why do you want
to claim to follow the spec? You can just make something that is very
similar to XMLHttpRequest but doesn't claim conformance, so users of
the API will know their stock XMLHttpRequest code may not work. The
whole point of specs is interoperability.
Regards,
Maciej
