- From: <Toman_Vojtech@emc.com>
- Date: Wed, 11 Mar 2009 12:41:33 -0400
- To: <xproc-dev@w3.org>
> I'd like to create an XProc pipeline to process this test document but > in order to do so I'd need a p:eval-pipeline step that had the ability > to take the pipeline declared in t:pipeline, and apply it to > the t:input > document. It could have the following signature: > > <p:declare-step type="p:eval-pipeline"> > <p:input port="source"/> > <p:input port="pipeline"/> > <p:input port="parameters" kind="parameter"/> > <p:output port="result"/> > </p:declare-step> > > All that I'd then need to do is use the p:compare step to > check that the > t:input document fragment matched the result from the p:eval-pipeline > step. > > I imagine that this feature would be a cinch for an EXProc extension > and, in the fullness of time, I'll be implementing it in my own XProc > processor 'Half-pipe'. I was thinking about something like this before, too. A step that takes an XProc pipeline and executes - that would be indeed a cool feature.. But the problem is that unless you restrict yourself to a specific class of pipelines (like what you did: p:pipeline with a single document input port, a single parameter input port, a single output port, no options), you probably can't really define the signature of the p:eval-pipeline step (or better: maybe you can, but it will probably look quite scary...). If we did what you are suggesting (support only certain classes of pipelines), my personal feeling is that it would be really annoying for the users to know that there is this great p:eval-pipeline step, but it just does not work for their pipelines (because they use options, or a document input port that takes a sequence of documents, or two output ports, or...) Just my two cents. Regards, Vojtech
Received on Wednesday, 11 March 2009 16:42:24 UTC