Re: [rest-discuss] RE: XSLT2.0 and HTTP-PUT

"Kay, Michael" wrote:
> Thanks for the comment. I think it's hard for us to be prescriptive about
> this. An XSLT processor designed solely for use on TV set-top-boxes might
> have no market requirement (or technical possibility) to write output to an
> HTTP PUT destination, nor indeed to anywhere other than the TV screen. We
> want language interoperability, but we can't assume that all processors
> execute in the same kind of environment.
> I don't mind, however, putting in some kind of note giving examples of the
> sorts of URI that permit writing.

Speaking for myself, I would be very happy with a non-normative note
that would get across the point (in appropriate spec-ese) that HTTP is
in fact a bidirectional protocol and that it would make just as much
sense to PUT to an HTTP URI as to a file system path. If at all
possible, please use the word "PUT" (if even in a non-normative example)
so that we have a chance of interoperability among implementations.
(there will be a tempation to use the more popular "POST" even though it
does not have the same "overwrite file" semantic).

Something like:

"For example writing to a file://... URI would overwrite a file. Writing
to an http://... URI would invoke the HTTP PUT method."

 Paul Prescod

Received on Tuesday, 30 April 2002 22:24:54 UTC