On 7 mai, 21:19, Daniel Marcel Eichler <[EMAIL PROTECTED]> wrote: > Am Dienstag 06 Mai 2008 16:07:01 schrieb Mike Driscoll: > > > If so, then it looks like an Interface is a generic class with method > > stubs. You can do that with Python just as easily by creating empty > > methods with just the "pass" keyword. > > Well, no. It's a litte different. Interfaces force to implement methods. > Simple inheritance with method-stubs can't garantee that a specific > method is reimplementat in a sub-class.
And Java's "Interface" wart can't garantee that a specific method is *appropriately* implemented. One often sees do-nothing methods in Java code - useless boilerplate code that's only here to satisfy the compiler. It's just like Java's "checked exception" mechanism : one very often sees do-nothing catch-all try/catch blocks in Java - which is way worse than just letting the exception propagate. I find all this totally pointless, because there's just no way for a compiler to check if your code is logically correct. > Interfaces work at > compile-time, while method-stubs raise at their first call, so in worst > case, never. And then ? If it's never called, why bother implementing it ? Here again, I don't see the point of compile-time checking. RuntimeError is a well-known Java exception, so the fact is that Java forces you into either rigid designs or overly complex workarounds (best knowns as "design patterns"), without even being able to garantee anything wrt/ correctness nor even safety. Not that static typing by itself is necessarily bad - as usual, there's a balance between somewhat conflicting goals -, but Java's type system definitivly is (bad). (snip) > That's the point. Interfaces garantee that a duck is a duck, an not only > a chicken that quack. Who cares if it's a chicken as long as it quacks when you ask her to ? Now *This* is the whole point of chicken^Mduck typing, isn't it ?-) -- http://mail.python.org/mailman/listinfo/python-list
