[oslc-core] "One last" change to OSLC Core representations

[Dave]

On Mon, Jul 26, 2010 at 9:41 AM, Samit Mehta <samit.mehta at us.ibm.com> wrote:
> I would recommend stronger language for the following paragraph:
>
>> OSLC domain specifications are expected to (1) require the
>> representations needed for the specific scenarios that they are
>> addressing...
>
> Maybe something like:
>
> "OSLC domain specifications MUST (1) require the representations ..."
>
> I do agree with your argument, but I am concerned that it could put a
> service consumer (client) at a disadvantage if a domain spec ends up
> without a firm requirement for at least one representation for each
> scenario.

Thanks for the feedback Samit.

Having a "firm requirement for at least one representation" does help
clients, but we are not there today in our implementations. Today,
OSLC implementations have and will continue to have for some time
limitations in their abilities to provide and accept RDF/XML. Changing
to SHOULD acknowledges fact and still points the way forward.

The term SHOULD is actually a pretty strong requirement, here's what it means:

SHOULD   This word, or the adjective "RECOMMENDED", mean that there
   may exist valid reasons in particular circumstances to ignore a
   particular item, but the full implications must be understood and
   carefully weighed before choosing a different course.

I think SHOULD is the right level of requirement for 2010. And, I
suspect that all OSLC implementations this year will support RDF/XML
to the best of their abilities -- so you will get what you want even
without the MUST.

Thanks,
- Dave