- From: Alex Milowski <alex@milowski.org>
- Date: Sun, 06 Aug 2006 12:51:26 -0700
- To: public-xml-processing-model-wg <public-xml-processing-model-wg@w3.org>
In writing the XML Schema for what we decided I found one input inconsistency that just bothers me. There are two cases where I think we have this inconsistency that I'll enumerate below. 1. Calculated pipeline inputs: We can now "calculate" a pipeline input by declaring another input that references a pipeline input and then uses a select to get a sequence of documents: <pipe name="ex1" ...> <declare-input port="main"/> <declare-input port="calculated" ref="ex1!main" select="//chap"/> </pipe> 2. for-each/viewport/etc. : <for-each name="iteration"> <declare-input port="chap" ref="somestep!output" select="//chap"/> </for-each> These are the only inconsistency we have with declare-input so far. In all other cases, declare-input just declares a name of a port available for binding. I suggest we make these what they are and just call them 'input'. We can continue to have declare-input for the current variety where only the name needs to be declared. In the above examples, we'd have: 1. Calculated pipeline inputs: <pipe name="ex1" ...> <declare-input port="main"/> <input port="calculated" ref="ex1!main" select="//chap"/> </pipe> Now the input is a calculated "input". 2. for-each/viewport/etc. : <for-each name="iteration"> <input port="chap" ref="somestep!output" select="//chap"/> </for-each> Here the interpretation would be we're calculating the input for the iteration based on the ref and select attributes. If we choose to collapse the 'declare-input' and 'input' names, we'll still need to distinguish between declarations of input names that need to be bound to some input and calculated inputs that are calculated against some existing available document. -- --Alex Milowski
Received on Sunday, 6 August 2006 19:58:59 UTC