Re: [cmis] Namespace proposal and complex properties with lists

[Julian Reschke <julian.reschke@greenbytes.de>]

Davis_Cornelia@emc.com wrote:
> I'd like to make sure that I'm understanding the current namespaces
> proposal correctly.  The proposal has two main parts:
> 
> 1.  Allow names (typeId and queryName for types and name for properties)
> to have prefixes

Yes.

> 2.  Define a new attribute for both types and properties called
> globallyUniqueName.

Yes.

These two are really orthogonal; so we probably could vote on them 
separately.

> The point of #1 is to scope names WITHIN a single repository allowing,
> for example, one type to have a property called foo that is distinct
> from a different type having a property called foo; the former could be
> defined as type1:foo and the latter type2:foo.  These prefixes do
> nothing to ensure uniqueness of property or type names across
> repositories because there is absolutely nothing to say that two
> repositories cannot both define a type1:foo.  Sound right? 
> 
> Of course, the globallyUniqueName is for just that - uniqueness across
> repositories.  But then we have to do a bit more specifying.  For
> example, how are properties to be identified in resource representations
> - via "name" or via "globallyUniqueName"?

My understanding was that not all implementations can provide 
globallyUniqueNames, thus it would be optional.

That being said, I think it would be good if it would appear on the 
wire, because otherwise clients will need to retrieve the property type 
information as well (or are they going to do that anyway?).

BR, Julian
-- 
<green/>bytes GmbH, Hafenweg 16, D-48155 Münster, Germany
Amtsgericht Münster: HRB5782