Re: A Strange Viewport

On 10/3/06, Alex Milowski <alex@milowski.org> wrote:
>
> Norman Walsh wrote:
> > / Alex Milowski <alex@milowski.org> was heard to say:
> > | Surprisingly, you can do this:
> > |
> > | <p:viewport name="crazy">
> > |    <p:declare-input port="e" step="other" source="result"
> > |                     select="/doc/x"/>
> > |    <p:declare-output port="replacement" step="crazy" source="e"/>
> > |
> > |    <p:step name="a" type="special">
> > |    <p:input port="document" step="outside" source="result"/>
> > |    </p:step>
> > |
> > | </p:viewport>
> > |
> > | In this case, the viewport iterates over the matching 'x' subtrees
> > | in the document but doesn't replace them.
> > |
> > | The contained step 'a' should be executed as many times as there
> > | are 'x' subtrees but it doesn't use as input anything in the
> > | viewport and doesn't contribute to the viewport's output.
> > |
> > | From and input/output graph flow, it isn't in connected to the viewport
> > | in anyway.
> > |
> > | Should we allow this kind of step in iteration (i.e. viewport or
> > | foreach)?
> >
> > I don't see why not. I don't even see how to prevent it if we wanted
> > to.
>
> One constraint that would make this invalid is:
>
> "Steps inside iterations must be traceable to the subject of the
>   iteration."
>

I think it could probably be a warning if the pipeline engine could
detect it (which in some narrow case wouldn't be so obvious).

But not allowing it could be a heavy load for implementors

Some implementation would detect it and make the simplification on the
fly if necessary

For me it is to relate to the same level of a step in a pipeline which
is unused. it could be just here for beeing used later.

Mohamed

-- 
Innovimax SARL
Consulting, Training & XML Development
9, impasse des Orteaux
75020 Paris
Tel : +33 8 72 475787
Fax : +33 1 4356 1746
http://www.innovimax.fr
RCS Paris 488.018.631
SARL au capital de 10.000 €

Received on Tuesday, 3 October 2006 12:33:06 UTC