Re: Extensibility of c:result

Thanks Vojtech for solving those parts of the question :
1) no model, I'm perfectly fine with that
2) propose c:error instead of c:result in case of error : that sounds good !


But it doesn't solve the extensibility question as to :
a) what if, as an implementer, I want to provide more info in this c:result
b) what if, in future version, we want to add more info into the c:result

Mohamed

On Tue, Apr 20, 2010 at 1:26 PM, <Toman_Vojtech@emc.com> wrote:

>  The content-model of c:result is overloaded already in the XProc
> specification itself: see the p:exec step. For text output with
> wrap-result-lines=true, it produces the following result:
>
>
>
> <c:result>
>
>   <c:line>line1</c:line>
>
>   <c:line>line2</c:line>
>
>   …
>
> </c:result>
>
>
>
> But I don't think the content model of c:result is defined anywhere in the
> spec (and I am not sure it is a good idea to do that. IMHO, it is just a
> wrapper with no special meaning). Section 7 (Standard Step Library) says
> that: "…several steps use this element for result information:
>
>
>
> <c:result>
>
>     string
>
> </c:result>
>
> "
>
>
>
> but I read this more as a hint of what to expect from a typical standard
> XProc step.
>
>
>
> I remember raising this long time ago but I think the conclusion of the WG
> was that this not really an issue.
>
>
>
> --
>
>
>
> For EXProc, my opinion is that we can just fix the problem of
> distinguishing between a successful/unsuccessful c:result easily by saying
> that if you use fail-on-error=false, you get an c:error document, and not a
> c:result document. Or something like that.
>
>
>
> Vojtech
>
>
>
>
>
> *From:* public-xml-processing-model-wg-request@w3.org [mailto:
> public-xml-processing-model-wg-request@w3.org] *On Behalf Of *Innovimax
> SARL
> *Sent:* Tuesday, April 20, 2010 1:05 PM
> *To:* XProc WG
> *Subject:* Extensibility of c:result
>
>
>
> Dear,
>
>
>
> I think this is the right place for that discussion
>
>
>
> We use for the moment c:result in many places in the spec, but the way it
> is used is not extensible
>
>
>
> The content is most of the time a single string as a direct child
>
>
>
> I know we moved on from attribute because it was hard to get the infomation
> from it, but could we better go for an extra wrapper ?
>
> <c:result>
>
>   <uri>http://....</uri>
>
> </c:result>
>
>
>
> I'm saying this because I already see some problem of using c:result for
> example when there is an error or not (in exproc pxf:copy) : we should be
> able to make the difference between those two
>
>
>
> Of course, we could expect that if we will need extension then we will add
> an extra attribute to c:result, like @version=2 telling that you will need
> special care about the content, but it will make existing pipeline non
> conformant
>
>
>
> The second point, is that a conformant implementation might need to output
> extra information using it's own namespace and that would not be possible
> (unless using attribute which is pretty limited)
>
>
>
> 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 €
>



-- 
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 €