Steven D. Majewski:
> However, I don't think 'setattr' with no value meaning to delete
> the attribute is a necessarily 'special case' in the *language*.
> I seems like a perfectly sensible semantics. I think what
> Guido meant was that it was a 'special case' for the *implementation*.
>
> But, I have no objection to 'delattr', either.
I guess Fred and others are right -- overloading delattr() on
setattr() is at least surprising. This is done at the C level, mostly
for historic reasons (and also because often there is a lot of common
checking for setattr and delattr), but there's no reason to let this
affect the interface to which Python users are exposed.
--Guido van Rossum, CWI, Amsterdam <Guido.van.Rossum@cwi.nl>
<URL:http://www.cwi.nl/cwi/people/Guido.van.Rossum.html>