homepage

It seems python 2.6.2 (at least under Windows, I have not tested other
platforms) does break the len function on the reversed iterator:
Python 2.5.2 (r252:60911, Feb 21 2008, 13:11:45) [MSC v.1310 32 bit
(Intel)] on
win32
Type "help", "copyright", "credits" or "license" for more information.
>>> len(reversed([1,2,3]))
3
Python 2.6.2 (r262:71605, Apr 14 2009, 22:40:02) [MSC v.1500 32 bit
(Intel)] on
win32
Type "help", "copyright", "credits" or "license" for more information.
>>> len(reversed([1,2,3]))
Traceback (most recent call last):
 File "<stdin>", line 1, in <module>
TypeError: object of type 'listreverseiterator' has no len()
I don't think it was meant to work but it did in Python 2.6.1 (and 2.5
as shown above). I guess it has something to do with Issue #3689 and
guess __len__ was simply removed. 
Problem really is that Python 2.6.2 breaks backwards compatibility.

Raymond, this was your change in r67478 (backported to trunk in r67498).

Guido decided that iterators should not support len() because he wanted
bool(it) to always be True.

Uh, perhaps the behaviour wasn't optimal but breaking compatibility
between two bugfix releases isn't developer-friendly either.
While we could keep it in trunk, it sounds like the change should be
reverted in 2.6.
The backport to 2.6 was done by Georg in r67569, by the way :-)

It's up to you guys. I had thought to change it only for Py2.7 but
Guido probably considers it to be a bug, so possibly the backport was
justified.

A compatibility break in a minor bugfix version is never a good idea I
guess. I understand the reasoning but this change may break quite a few
packages (at least it broke mine). A workaround in programs is easy but
for already released and deployed versions simply not possible. A change
for Python 2.7 would be at least annoying too but I guess could be
justified. Just my thoughts...

Is there a 2.6 release imminent? I thought I saw some discussion of a 
2.6.3 release, but I'm not sure what the eventual decision was. If so, 
perhaps this change could be quickly reverted in the release26-maint 
branch?

Setting as a release blocker so Barry can consider this for 2.6.3
(-committers seems to suggest that it'll be released soon).

Essentially, the only argument for reverting this is breaking
compatibility with 2.6.0 and 2.6.1. But, in the process, reverting it
means breaking compatibility with 2.6.2. Since the cat is already out
of the bag for 2.6.2, I think the bugfix ought to be left in-place. 
It's too disruptive to switch back and forth during micro releases.

If 2.6.3 isn't imminent, I agree this change should be left in place.

Is it okay with everyone to close this as "wont fix"? (BTW, is there
some reason that the Resolution field isn't allowed to have an
apostrophe in it?) There's nothing we can do about 2.6.2; it appears
2.6.3 wasn't imminent, and as Raymond says it seems disruptive to
switch it back now.

> Is it okay with everyone to close this as "wont fix"? (BTW, is there
> some reason that the Resolution field isn't allowed to have an
> apostrophe in it?) There's nothing we can do about 2.6.2; it appears
> 2.6.3 wasn't imminent, and as Raymond says it seems disruptive to
> switch it back now.
Well, the harm is done, unfortunately, and both resolutions have
analogous downsides, so we might indeed close it.

Oh, answering by e-mail reopened the bug for some unknown reason.

Closing.