- From: Innovimax SARL <innovimax@gmail.com>
- Date: Thu, 9 Aug 2007 21:24:25 +0200
- To: "Norman Walsh" <ndw@nwalsh.com>
- Cc: public-xml-processing-model-wg@w3.org
On 8/9/07, Norman Walsh <ndw@nwalsh.com> wrote: > Also on the 9 Aug 2007 call, Mohamed asked us to reconsider our > decision to support defaulted input and output ports on p:pipeline so > that this sort of pipeline is valid: > > <p:pipeline xmlns:p="..."> > <p:identity/> > <p:pipeline> > > We didn't get to that item either, but I'm anxious to see what we can > closed. > > We did discuss this in a previous call and the status quo is that this > is legal. > > Does anyone else want to revisit that decision? > > (Mohamed, this is your queue to explain why you think we've made a > mistake ;-) > Well, I take on :) I think that even if the idea was appealing, the big business of implicit inputs and outputs looks to me as a V.next feature I would like at least that for V1, the pipeline explicitly declares its inputs and ouputs (note that I'm talking about I/O of the pipeline only) Let's give some troubling example <p:pipeline> <p:xslt/> </p:pipeline> what is supposed this one to be equivalent to ? How many implicit inputs are generated ? Now that we have parameter ports it seems more and more unlikely to be fully manageable by users (may I recall than nothing forbid an input of kind="document" to be connected to an input of kind="parameter" And what about : <p:pipeline> <p:xslt/> <p:split-sequence test="position() mod 2 = 1"/> </p:pipeline> I think that such shortcut seem very appealing but are going to make verification very tortuous Mohamed -- Innovimax SARL Consulting, Training & XML Development 9, impasse des Orteaux 75020 Paris Tel : +33 9 52 475787 Fax : +33 1 4356 1746 http://www.innovimax.fr RCS Paris 488.018.631 SARL au capital de 10.000 €
Received on Thursday, 9 August 2007 19:24:28 UTC