- From: Alessandro Vernet <avernet@orbeon.com>
- Date: Thu, 3 May 2007 16:04:50 -0700
- To: public-xml-processing-model-wg <public-xml-processing-model-wg@w3.org>
On 5/3/07, Henry S. Thompson <ht@inf.ed.ac.uk> wrote: > We seem to be converging on the > > option:parameter::pre-known:not-pre-known Yes, but we need to say more about what "pre-known" means. Consider the case of XSLT. Parameters are not pre-known to the XSLT component. But they are pre-known to the XSLT stylesheet, and in most cases are pre-known by author of the pipeline that calls the stylesheet. Bare with me, this matters for what follows. > 3) No more parameter importing -- it likewise serves no purpose -- > all in-scope parameters are available to a step. I don't like passing all parameters in scope by default, as I mentioned during our call today. Passing all the parameters in scope is useful when the pipeline author doesn't know what parameters a stylesheet takes, and calls the stylesheet assuming that whoever will the pipeline will know what the stylesheet needs and will pass the right parameters to the pipeline. If I remember correctly Norm mentioned some cases where this would be desirable when using DocBook. But personally, this is a use case I have never seen. I think that the 99% case is that when pipeline authors call a stylesheet they know what parameters that stylesheet takes, and they can bind those parameters explicitly in the step. So passing everything in scope is not needed in the 99% case; let's not do that by default. Going one step further: given the lack of use cases asking for all parameters in scope to be passed to a step, why don't we keep things simple, and make parameters binding alway explicit? Alex -- Orbeon Forms - Web 2.0 Forms for the Enterprise http://www.orbeon.com/
Received on Thursday, 3 May 2007 23:04:54 UTC