[Zope-CMF] Re: Plone participation in the CMF list

[Rob Miller]

Florent Guillaume wrote:

> Tres Seaver  <tseaver at palladion.com> wrote:
> 
>>I think the discussion around Archetypes, in particular, ended up
>>stalled over the question of whether to "code generation" design
>>should be preferred over "configuration-based" design (as found in
>>CPSSchemas, for instance).
> 
> Also now that Zope 3 is taking more and more importance in CMF, any
> schema-based solution should be based on Zope 3 schemas. IMO both
> Archetypes and CPSSchemas are too big frameworks to include in CMF.

maybe i'm being naive here, but what's wrong w/ trying to get them all 
to play together?  i've been doing work on ATSchemaEditor and have been 
thinking about ways to bring it forward.  the first obvious choice is to 
use adapters to designate an object as a schema consumer 
(ISchemaConsumer), able to be served up its schema from a schema 
provider (ISchemaProvider).  even just within the AT world, there is the 
problem of schemas that will be partly defined via python code, and 
partly defined in blob space via a TTW schema editor.  the use of an 
ISchemaDefiner interface could possibly be used as a bridge between any 
schema definition framework (AT, Z3, CPS, XML, blob-space, etc.) and the 
schema providing layer.

of course, some things are bound to get lost in translation, and there 
will probably be some strongly held (and differing) opinions on what, 
exactly, a ready-to-be-consumed schema should specify, but an 
abstraction layer like this in CMF would at least make it easier for the 
rest of us to plug in.

-r