- From: Alessandro Vernet <avernet@orbeon.com>
- Date: Thu, 13 Dec 2007 08:08:26 -0800
- To: public-xml-processing-model-wg <public-xml-processing-model-wg@w3.org>
On Dec 13, 2007 2:21 AM, Henry S. Thompson <ht@inf.ed.ac.uk> wrote: > I wasn't suggesting it _would_ be allowed. My proposal is to make > p:pipeline a convenience for the 90% case: it effectively corresponds > to a p:pipeline-library containing a single p:declare-step with > attributes and contents taken from the p:pipeline. That's just a way > of thinking about, or defining its semantics -- it's up to the editor > whether, if we adopt this in principle, he describes it that way, or > in its own terms. Right, if we go with the p:declare-step / p:pipeline proposal, it would certainly be up to the editor to decide if he wants to describe one in term of a special case of the other. Understood. With this in mind, I was wondering why we would need 2 distinct constructs: can't we have just one construct, which can be used both in a p:pipeline-library and on its own (root element)? The benefit is two-fold: you simplify the language by having only one construct instead of two, and more importantly you make every single pipeline a reusable piece of code, just as if it was declared in a pipeline library. Alex -- Orbeon Forms - Web 2.0 Forms, open-source, for the Enterprise http://www.orbeon.com/
Received on Thursday, 13 December 2007 16:08:38 UTC