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

Re: p:pipeline

From: Jeni Tennison <jeni@jenitennison.com>
Date: Thu, 27 Jul 2006 09:50:41 +0100
Message-ID: <44C87E61.8080704@jenitennison.com>
To: public-xml-processing-model-wg <public-xml-processing-model-wg@w3.org>


Alessandro Vernet wrote:
> Unlike inner pipeline, there a number of use cases where calling a
> pipeline from a pipeline is needed. But we don't necessarily need to
> add something to the language for that. Arguably, it would even be
> more elegant not add something to the language. Why would we have a
> special construct to call a pipeline, vs. say an XSLT stylesheet.
> Running a pipeline could be done by invoking a 'pipeline' component in
> a step, just like to run a stylesheet we invoke an 'xslt' component in
> a step.

(We've discussed this before.) Two reasons for not doing having a 
'pipeline' component:

1. It means pipeline definitions can be dynamically generated by other 
pipelines, which means that pipelines can't be compiled.

2. Calling a pipeline becomes very difficult. The 'call-pipeline' 
component needs to have a 'pipeline' input, an 'inputs' input, and an 
'outputs' output. All the inputs for the pipeline have to be defined 
somehow within the 'inputs' input, which requires another complex 
syntax. This is analogous to the problem we have with passing XSLT 
parameters to an XSLT component.

Of course processors are free to define such a component, but it 
shouldn't be the normal way of invoking (static) pipelines.


Jeni Tennison
Received on Thursday, 27 July 2006 09:17:49 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:32:40 UTC