Re: Minutes for XProc WG telcon of 6 Apr 2006

Norman Walsh wrote:
> / Rui Lopes <rlopes@di.fc.ul.pt> was heard to say:
> | Well, if we want consistency between steps and pipelines, I believe we should
> | allow the definition of simple parameters (i.e. booleans, integers, etc.) in a
> | pipeline, the same way it has been talked regarding steps parameters. Having
> | this helps on defining a standardized sub-pipeline processing component.
> 
> I'm not sure about booleans and integers. I think I'd like to keep
> types out of it and just let parameters all be strings.

Yes, strings should be sufficient for "simple" parameters. A least for 
XProc 1.0.

> I was thinking we could perhaps harmonize the two by saying that the
> pipeline parameters become variable bindings in the static context for
> XPath evaluation of p:if component expressions. Then we could say that
> a p:if component without any test document evaluates its expression
> against an empty document node.

This was what I had in mind. It works similarly to XSLT (which I believe 
potential XProc hackers are used to work with).


>       <p:save href="someURI.html">
>         <p:input name="document" label="doc"/>
>       </p:save>

Do we benefit on allowing defining a mime-type attribute on p:save and 
p:load elements, hinting how to (de)serialize documents? If we do so, 
it's another argument for p:load's existence.


Cheers,
Rui