On Apr 16, 10:40 am, Aaron Watters <[EMAIL PROTECTED]> wrote:
> On Apr 16, 12:27 pm, Rhamphoryncus <[EMAIL PROTECTED]> wrote:
>
> > On Apr 16, 6:56 am, Aaron Watters <[EMAIL PROTECTED]> wrote:
>
> > > I don't get it.  It ain't broke.  Don't fix it.
>
> > So how would you have done the old-style class to new-style class
> > transition?
>
> I'd ignore it.  I never understood it and never had
> any need for it anyway.  New-style classes and metaclasses
> were a complicated solution to an unimportant problem in
> my opinion.  And also a fiendish way to make code
> inscrutible -- which I thought was more of a Perl thing
> than a Python thing, or should be.
>
> I must be missing some of the deeper issues here.  Please
> educate me.

>>> type(3)
<type 'int'>
>>> (3).__class__
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
AttributeError: 'int' object has no attribute '__class__'
>>> class Foo: pass
...
>>> type(Foo())
<type 'instance'>
>>> Foo().__class__
<class __main__.Foo at 0x811f47c>

With new-style classes, (3).__class__ returns int and type(Foo())
returns Foo.

Of course a lot of other aspects were redesigned at the same time,
such as how attributes/methods are looked up.  The consequence of that
is we got descriptors.  Seems like a fair trade to me.

Another example of an incompatible change was this:
>>> 2**32
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
OverflowError: integer exponentiation

In recent versions of 2.x you get this instead:
>>> 2**32
4294967296L

Finally, 3.0 changes it to this:
>>> 2**32
4294967296

Personally, I find all these changes to be a good thing.  They make
the language cleaner and more useful.  The only reason to not make the
changes is that old, crufty, unmaintained libraries & applications
might depend on them somehow.  If that's more important to you, what
you really want is a language who's specs are frozen - much like C
effectively is.  I hope python doesn't become that for a long time
yet, as there's too much it could do better.
-- 
http://mail.python.org/mailman/listinfo/python-list

Reply via email to