Re: Cardinality of inputs

On 6/7/07, Norman Walsh <ndw@nwalsh.com> wrote:
> Right now, we have support for 1 or 0-or-more documents as input.
> What's missing is support for 1-or-more and 0-or-1.
>
> I can think of some reasons for 0-or-1 (an optional configuration
> input, for example), but I can't think of any uses for 1-or-more.

0-or-1 is already in use in the spec : look at p:option/p:parameter
but since zero is authorized, why an empty sequence wouldn't be
authorized here ?


for 1-or-more, I agree that's a bit of sequence where you mandate a
content, and I also don't catch a serious use case


>
> But we don't have any atomic steps which need the missing options, so
> we're considering a change that would only be used by some future (or
> extension) atomic steps or by pipeline authors.
>
> For pipeline authors, it seems that you'd have to handle the
> optionality anyway, so handling the case of more than 1 when you
> expected 0 or 1 wouldn't be a burden.
>
> Absent more compelling arguments for changing the status quo, I don't
> think any change is likely to be made.
>
> If we do make a change, there was clear sympathy for a position which
> says that neither of the two existing cases should be made more
> difficult. That is, exactly 1 should be the default and 0-or-more
> should require at most one attribute.
>
> With that in mind, does anyone want to advance arguments for change
> in this part of the spec?

some DTD like

(nothing) or cardinality=""      1-1
cardinality="*"                      0-or-more
cardinality="?"                      0-or-1
cardinality="+"                     1-or-more



-- 
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 Thursday, 21 June 2007 07:43:22 UTC