[Zope3-dev] Context and Component Lookup

[Jim Fulton]

Steve Alexander wrote:
> 
>> There is no formal way to document this in either Zope or PEAK at 
>> present; it's just "documentation".  In PEAK, though, components have 
>> explicit "out-bindings" like the pins of a chip, so you can usually 
>> browse these in the source without reading method bodies to see what 
>> lookups are occurring.  (Which is probably why I didn't think about it 
>> explicitly in this discussion.)  Perhaps a similar approach would be 
>> useful in Zope, or perhaps some type of "component requirements 
>> schema" object, if that's more fitting of Zope's overall style.
> 
> 
> I'm *particularly* keen on something like this.

Me too, though perhaps not urgently so.


> Many researchers in CBSE would claim that Zope 3's CA is not really a 
> "component" architecture because an important aspect of something being 
> a component is that you can inspect its interfaces and its dependencies.
> 
> You should be able to infer from inspecting a component and looking at 
> the system whether a component's dependences are met.

I agree. I'd lovee to see some proposals and contributions in this area
at some point.  To some degree, I think we've been at too early a stage in
the development of our ideas to attack this, but perhaps the time is
coming soon.

For example, I think that while we've been getting started, we've needed to
do what it took to get things working. We has a bit of an explosion of different
component types.  Adaptergeddon will unify most component types into various
dimension of named adapter.  I think that this will make it easier to reason
about component dependencies.

Jim

-- 
Jim Fulton           mailto:jim at zope.com       Python Powered!
CTO                  (703) 361-1714            http://www.python.org
Zope Corporation     http://www.zope.com       http://www.zope.org