Re: OOP and Relationship to the Guile Module System

[Maciej Stachowiak <mstachow@mit.edu>]

russell.mcmanus@gs.com writes:
> Klaus Schilling <Klaus.Schilling@home.ivm.de> writes:
> 
> > Mathematical set theory and logics has to be careful about avoiding antinomies
> > like those of Russel, Burali-Forte and Cantor, all caused by meta-sets. 
> > Zermelo's set axioms eliminate the danger of antinomies.
> >  
> >  May an improperly designed meta-stuff protocol also lead to antinomies?
> > How are they avoided in CLOS ?  
> 
> There is Kiczales' (sp?) 'The Art of the MetaObject Protocol', which
> probably has the answer.  This seems like a hard question to answer on
> a mailing list.

I don't think the paper will address it because it's a rather obscure
concern. All paradoxical sets that existed under pre-Zermelo-Frankel
set theory are not recursively enumerable, and are therefore unlikely
to be a concern for a programming langauge (unless the language is
computationally stronger than a Turning machine - which is not
likely).

More generally, an OOP class bears only a superficial resemblance to a
set.

There is no way, for instance, to define "the class of all class
objects that are not instances of themselves" or anything like
that.

 - Maciej

PS I apologize for increasing the level of obscure math on the list.