W3C home > Mailing lists > Public > public-xml-processing-model-wg@w3.org > June 2006

Re: Hiding dependencies in XPaths (Was: XProc Minutes 25 May 2006)

From: Richard Tobin <richard@inf.ed.ac.uk>
Date: Thu, 1 Jun 2006 17:18:14 +0100 (BST)
To: Jeni Tennison <jeni@jenitennison.com>, public-xml-processing-model-wg@w3.org
Message-Id: <20060601161814.625AB6D8703@macintosh.inf.ed.ac.uk>

I think we may have a misunderstanding here.  I was considering (and
rejecting) something like this:

<p:pipeline>
  ...
  <p:step name="foo" type="whatever">
    <p:output name="output" label="foo-output"/>
    ...
  </p:step>
  ...
  <p:if test="$foo-output/name = 'smith'">
    ...
  </p:if>
  ...

where the XPath in the conditional is referring to an output of some
other step in the pipeline by using its label as a variable name.  This
means that the dependency of the conditional step on the "whatever"
step is only determinable by inspecting the XPath.  I think that's bad.

But if the idea is something like this:

  <p:if test="$test-input/name = 'smith'">
    <p:input name="test-input" ref="foo-output"/>
    ...

where the variables refer only to the step inputs, then I have no
objection.  The dependency is clear from the inputs.

A simple extension would be to say that the first input is available as
the context node, so that you don't have to use a variable if there is
only one input.

-- Richard
Received on Thursday, 1 June 2006 16:18:23 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:21:47 GMT