Copy constructors

Guido van Rossum guido at python.org
Mon Aug 13 23:03:30 EDT 2001


Glyph, give it a rest.  I hear loud and clear what you are saying.  At
this point, stepping up the rhetoric will have an adverse effect on
your case.  We may always disagree in how important we find the
various aspects of Python, but I know what you think is important.
Please understand what I think is important, even if you disagree -- I
don't expect to change your mind, so please do me the same favor and
stop trying to change my mind.

That said, the decisions I am making for the new type system in 2.2
are not cast in stone.  I consider the new class mechanisms added to
2.2 the first experimental release of a very far-reaching new feature.

Some important things to keep in mind:

(a) Your existing dynamic classes won't break in 2.2, *unless* you
    explicitly choose to enable the new class mechanisms (e.g. by
    inheriting from 'object' or using an explicit __metaclass__).

(b) Based on the experiences with use of the new class mechanisms as
    deployed in 2.2, I'll refine the design for 2.3.

(c) I expect that "classic" classes will remain the default until
    Python 3.0, where they will disappear -- by then, almost everybody
    will be using the new classes, though we may have a backwards
    compatible metaclass for code that connot be converted completely.

--Guido van Rossum (home page: http://www.python.org/~guido/)



More information about the Python-list mailing list