> Imho, there are already too many meta-models in the squeak
> world, just as there are too many event frameworks. OB has one
> that defines what properties a node in a tree can take on.
> FileList has one that describes what actions can be done to a
> file. Preferences has one. Etoys had a huge one, but let's not
> even go there.
> 
> I don't know if it belongs in the core (I don't know what the
> core is), but it is a basic service I'd expect a lot of
> libraries could depend on.
> 
> --
> Matthew Fulmer -- http://mtfulmer.wordpress.com/

Choice is not a bad thing, there is no uber meta model that will suite
everyone's needs all of the time.  There's nothing wrong with projects
creating their own meta model that is suited to exactly what they need.  

For most things, I find pragmas more than sufficient for static metadata and
much lighter weight than Magritte both in code and brain power.  There's
also the nice synergy of having your metadata actually bound to the things
they're representing.  This is an advantage pragmas have over Magritte.

Magritte has the advantage when the metadata starts to become dynamic but
you pay for using it with a much heavier weight code centric approach that
separates the metadata from what it's representing.  Now you have to
maintain a link between the "thing" and the "meta thing" either by either
naming patterns or by using pragmas to bind them.  These are serious
trade-offs to make, it's certainly not a no brainer.

Ramon Leon
http://onsmalltalk.com


_______________________________________________
Pharo-project mailing list
[email protected]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

Reply via email to