Head on MagmaCollection

[Hilaire Fernandes]

Hi!

It is fun to experiment with Magma and persistency. Thanks Chris  for 
this great piece of software.

I would like to use it seriously in a pet project. It is a kind of 
e-learning application composed of two blocks: a local morphic 
application with pedagogical activites and a distant data base 
repository. The two applications could be on different network location 
on the Internet.

The data base repository contains knowledge representations and a very 
sharp log of the learner activity. So bascily after each learning act on 
the morphic application, a batch of log is saved on the distant 
database. The learner log is a tree populated with different Smalltlak 
objects (learningConcept, moprh, numbers, stream,...).

So far I have implemented all the persistency with the use of standard 
Smalltalk collection. Now I am thinking about optimizing the data flow 
between the morphic application and the database. In particulary when a 
learner log need to be recorded in the data base. As implemented right 
now -- just standard collection and no readStrategy -- I guess that 
inserting a new log in the learner log tree imply first fetching a lot 
of the tree locally then adding the new entries. It is not wise and it 
is bit slow. Now I am wondering if the MagmaCollection is something that 
could help.
Will it be possible to do a distant querry on the data base server to 
fetch the exact object from where I need to hook my new learner log?

Of course I could used readStrategy to fully load locally major part of 
the learner log tree -- and probably I will do that -- but I am currious 
about the MagmaCollection options.

Hilaire