[PN/NH] Component definition

Niclas Hedhman niclas at hedhman.org
Mon Nov 1 02:17:11 EST 2004

[ PN/NH stands for a series of mails reporting on the various discussions that 
I and Peter Neubauer have been talking about in face-to-face 'meetings' over 
the last few days on his visit to Kuala Lumpur. ]

Our current blocks can expose services, and later when we get to configuration 
routing, they are effectively components. But blocks are stand-alone entities 
that can be loaded and unloaded by themselves, which "native components" can 
not (for instance, may require classes not available in the classloader).

And if components are not loadable in a running server, the question could 
then be "Are they really Components, or merely building blocks or parts for 
the making of components?"

I would *REALLY* love to have a definition where a component is a loadable 
item into a running instance of a server. Today we call that a block. And the 
<component> would then be a "part".
Now, this imposes a change that I think is too great at this point, but 
perhaps it could be tackled by higher level and documentation definition 

"Block Oriented Programming"
"Block" with capital B, for the Metro definition of Block, just like 
"Component" was different than "component" in Avalon.
And we should probably structure the documentation along the lines of Block vs 
Part concerns, and so forth.


  / http://www.bali.ac        /
 / http://niclas.hedhman.org / 

More information about the dev-dpml mailing list