Re: Clarifying p:document, p:load wrt p:http-request

"Toman, Vojtech" <vojtech.toman@emc.com> writes:
> I like the direction of supporting multipart in p:load and allowing
> sequences on the output port. Especially the latter makes the step so
> much more flexible. Would it make sense to generalize the support for
> "compound types" beyond multipart and allow (probably as an
> implementation-defined feature) p:load to read other compound
> structures as well? For instance, I can imagine an implementation
> reading from a ZIP archive in the presence of the "zip" URI scheme
> etc.

If there's a zip: URI scheme, then I think the implementation is free
to do that under the "number and nature of URI schemes supported is
implementation-defined" rule.

> Regarding multipart, what about p:document? What do you get in this
> case? An error or a sequence of documents?

I'm proposing that p:document can also produce a sequence of documents:

  The semantics of p:document are the same as a the semantics of
  p:load where the href option comes from the href attribute, the
  override-content-type option comes from the override-content-type
  attribute, and the dtd-validate option is always false.

> Regarding globbing, I like the idea. Is there something comparable in XSLT?

Not that I'm aware of, though I believe Saxon supports an extension. I
think XML Calabash does too, actually.

                                        Be seeing you,
                                          norm

-- 
Norman Walsh
Lead Engineer
MarkLogic Corporation
Phone: +1 512 761 6676
www.marklogic.com