On Wed, 20 Jun 2007, Frank Barknecht wrote:

...

You mean like http://en.wikipedia.org/wiki/Decorator_pattern ? (original page at http://www.c2.com/cgi/wiki?DecoratorPattern )

Yes, I meant it as a reference to that design pattern, but more to the problem it tries to solve

Well, IMHO, a pattern is as much the problem being solved as it is the solution to the problem. It takes the problem to think of the solution related to it. (If it wasn't solving a problem, it wouldn't be a solution)

i.e. adding functionality to existing objects and not as much to the proposed solution i.e. adding that behaviour at runtime. I guess I'm proposing simple "subclassing" or wrapper-abstractions as solution.

I think that at the level of making pd abstractions, "subclassing" and "wrapping" can not be thoroughly distinguished.

If there is any feature that I'd want from inheritance that can't be done by simple wrapping, it would be linearisation of inheritance, including elimination of redundant base classes in the hierarchy. E.g. if you have an abstraction called [D] which wraps [B] and [C], and then both [B] and [C] each wrap [A], but you want only one instance of [A] to exist in total, then what do you do? This is the problem that linearisation of inheritance solves.

But actually I'm trying to not to propose anything at all at this stage.

why?

_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal QC Canada