Noel J. Bergman wrote:

The casting I was refering to was the "type of locator"

Why do we need more than one TYPE (interface)?  Define a consistent,
uniform, SINGLE, interface.  Aggregate by reference and naming space, as we
discussed previously with respect to URNs.


I would aggregate the two interfaces like this:
   interface MyLocator extends SystemLocator, MailetLocator {}

Me too.

Oiy ... please don't make me do it.  Please don't make me explain about IS-A
and HAS-A again.

Just a note - IS-A can define the HAS-A contact (providing meta is provided against the IS-A interface declaring the set of entries).

A Component should see a single "locator" object that provides access to
other objects.

As your reading though this stuff you will see that I'm convergin on the same opinion - one locator is sufficient.

I.e. the current:

<type>
<context type="o.a.a.p.BlockContext">
<entry key="apps.dir" type="java.io.File"/>
<!-- etc. -->
</context>
</type>

Can be replaced by:

<type>
<context locator="o.a.a.p.BlockContext" version="2.1"/>
</type>


The "meta-model" should provide the information necessary to register what
is in the context.  And there should be APIs for dynamic registation.

Agreed.

IMO, a Context should be a provider of objects within a given, well,
"context."  Those objects provide data and/or services.  Some of those
objects are well-defined and provided by the Container profiles.  Other
objects are defined by third parties, and usable primarily by others
expecting those third party entries.  The preceeding does not mean that a
third party can't replace a profile defined object, given the correct
administrative rights.

Yep.

Cheers, Steve.

	--- Noel


--
To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>



--

Stephen J. McConnell

OSM SARL
digital products for a global economy
mailto:[EMAIL PROTECTED]
http://www.osm.net




--
To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>

Reply via email to