Florent Guillaume wrote:
> In most uses cases I have, an empty list can be modeled by having the
> property not set. This is enough for me, but I agree that the JCR model
> for instance is more complex than that.
>
> I guess many existing repositories just don't support empty lists as a
> different value than null.
Well, some do (such as SAP KM, predating JCR, and every conforming JCR
implementation). Some don't.
> There's no easy solution, one model is richer than the other.
>
> In any case I wouldn't want to see this be an optional behavior. The
> spec should specify one or the other.
I agree that JCR's behavior is harder to implement. The main reason to
have it seems to be that you can make a multivalued property mandatory,
and still allow it to be empty. It probably doesn't make a big
difference in practice...
Al Brown wrote:
> ...
> I'd like to keep this simple and unless the majority of repositories
> support not set, set: null/empty, and set: value. Right now CMIS does
> not have a notion of null only set and not set. I'd also like to
> understand better what set: null/empty means compared to not set.
> ...
Well, exactly what the JCR spec says (it's quite clear on that, methinks).
(Keep in mind that we're talking about multivalued properties only).
BR, Julian
--
<green/>bytes GmbH, Hafenweg 16, D-48155 Münster, Germany
Amtsgericht Münster: HRB5782
