Re: Make p:http-request required?

/ Jeni Tennison <jeni@jenitennison.com> was heard to say:
| Why don't we? What makes it different from the p:http-request step?

Two things: First, the *response* from the http-request has to be a
structured c:http-response. So making the request a structured
c:http-request seems symetric.

My second reason hinges on the answer to a question I asked earlier
about what multiple c:body elements mean.

Right now, the p:http-request step accepts some things that could
obviously be p:options:

  method
  href
  status-only
  override-mimetype

It also accepts headers which could be p:parameters.

And it accepts a c:entity which could be an input except that it
contains zero or more c:body elements which are pairs of content types
and bits. I don't think we have a way of representing the c:body
elements without a wrapper.

If a wrapper is needed anyway, is it wrong to use the wrapper for
other things?

| It seems that we're making inconsistent design decisions across
| different steps. Both learnability and usability are greatly enhanced
| by consistency.

I see your point.

I accepted Alex's original design without really considering how else
it might have been designed. Perhaps your alternate suggestion is
really better from a consistency point of view:

<p:http-request>
  <p:input port="source" sequence="yes"/>
  <p:output port="result"/>
  <p:option name="method" value="post"/>
  <p:option name="href" required="yes"/>
  <p:option name="status-only"/>
  <p:option name="override-mimetype"/>
</p:http-request>

Where headers are passed as parameters (and we do have to revisit our
decisions about parameters).

What appears on the source port is the sequence of entity bodies. In
each case, if the root element of a body is a c:body element, then
the content type attribute is used to identify the content type and
the children of the c:body are sent as the actual body. If the root
element is anything else, then the content-type is assumed to be
application/xml and the entire document is sent as the actual body.

But I think you still need to get a c:http-response back.

                                        Be seeing you,
                                          norm

-- 
Norman Walsh <ndw@nwalsh.com> | Childishness follows us all the days of
http://nwalsh.com/            | our life. If anybody seems wise it is
                              | only because his follies are in keeping
                              | with his age and circumstances.--La
                              | Rochefoucauld