John J. Lee wrote: > The fact that "open classes" are apparently thought to be a good thing > in Ruby puzzles (and worries) me.
This objection strikes me as having the same nature as, "Python's lack of strong protection for class members puzzles (and worries) me". The Pythonic answer to that objection is usually that this is a feature: it lets people who know what they're doing solve problems more easily than if they had to work around a bunch of "helpful" protection. Classes are effectively open in Python, too, at least where methods are concerned, since one can do klass.__dict__["myMethod"]=myMethod. Though admittedly, it isn't widely advertised as a feature of Python. It makes sense to me to think of open classes as being open for extension, not modification. After all, methods added by clients are unlikely to depend on implementation details (though I suppose they could, to which I would say, "Don't do that"). Steve Yegge's "Opinionated Elf" is an example of a problem that is very easy and elegant to solve with open classes, and painful to solve when classes are closed: http://www.cabochon.com/~stevey/blog-rants/polymorphism-fails.html Cheers, -- JK -- http://mail.python.org/mailman/listinfo/python-list
