> From: Torsten.Bergmann at phaidros.com

>> We need the language independence and power of COM, the facilities of 
> CORBA and
> the platform independence of Java. Maybe Squeak will be the right
> platform
> but we have to do a lot for this. We need exchangable components in 
> Squeak
> which are independent from implementation. Mixing code by loading
> different
> change sets wont be the right way.

I absolutely agree with this:  Smalltalk/Squeak has the potential to  
be the ultimate 'glue' for hooking up components (think 'ma'!).  To  
achieve this, it must drop insisting on a 'closed world' where  
everything must be written in Smalltalk or hacked into its VM  
somehow.  Even the weaker assumption that 'everything is an object'  
is probably too strong.

Any object or computational component, written in any language,  
should be accessible as a first class peer within Squeak as long as  
can present some kind of proper interface, and adapters for that sort  
of interface can be written.

The Vitruvius group at CMU is doing some really, really interesting  
research in this area, from high-level descriptions of component and  
connector class to new work on separating computational functionality  
from its "packaging".

See for example the article on "Flexible Packaging".

http://www.cs.cmu.edu/afs/cs.cmu.edu/project/vit/www/paper_abstracts/DeLine.FlexPack.html

The ideas presented there are extremely powerful, and they require  
co-routines ( asynchronous messaging anyone? ) and stream-oriented  
processing to work well.  It is almost sad to think how much more  
elegant their approach would look in an async Smalltalk environment  
instead of the hacked-up C they use.

Marcel