Sven Marnach s...@marnach.net added the comment:
The behaviour discussed in this thread does not seem to be reflected in
Python's documentation. The documentation of __eq__() [1] doesn't mention that
objects should compare equal to themselves.
[1]:
Changes by Terry J. Reedy tjre...@udel.edu:
--
resolution: - fixed
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue4296
___
___
Python-bugs-list
Changes by Mark Dickinson [EMAIL PROTECTED]:
Removed file: http://bugs.python.org/file11989/issue4296.patch
___
Python tracker [EMAIL PROTECTED]
http://bugs.python.org/issue4296
___
Mark Dickinson [EMAIL PROTECTED] added the comment:
Before committing, please add one
other test that verifies the relationship between in in membership
tests and in in a for-loop:
Added. (To test_contains, since this seems like a more appropriate place
than test_float.)
Added file:
Christian Heimes [EMAIL PROTECTED] added the comment:
Mark, would you do the honor, please?
___
Python tracker [EMAIL PROTECTED]
http://bugs.python.org/issue4296
___
___
Raymond Hettinger [EMAIL PROTECTED] added the comment:
Reviewed the updated patch and all is well. Thanks.
___
Python tracker [EMAIL PROTECTED]
http://bugs.python.org/issue4296
___
Mark Dickinson [EMAIL PROTECTED] added the comment:
Committed, r67204. Thanks guys, and thanks especially to Michael for
spotting this before 3.0 final.
--
status: open - closed
___
Python tracker [EMAIL PROTECTED]
http://bugs.python.org/issue4296
Mark Dickinson [EMAIL PROTECTED] added the comment:
This is indeed interesting. For what it's worth, I think
the Python 3.0 behaviour is the right one, here. Perhaps
it's worth adding a test to Python 3.0 to make sure that
this behaviour doesn't accidentally disappear, or at least
to make sure
Mark Dickinson [EMAIL PROTECTED] added the comment:
[Raymond]
assuming basic invariants. In 2.6, all of the following are always
true:
assert a in [a]
assert a in (a,)
assert a in set([a])
assert [a].count(a) == 1
And these are all still true in 3.0 as well, aren't they?
In any
Raymond Hettinger [EMAIL PROTECTED] added the comment:
To be clear, I'm saying that PyObject_RichCompareBool() needs to
add-back the code:
/* Quick result when objects are the same.
Guarantees that identity implies equality. */
if (v == w) {
if (op ==
Michael B Curtis [EMAIL PROTECTED] added the comment:
All,
Thank you for your rigorous analysis of this bug. To answer the
question of the impact of this bug: the real issue that caused problems
for our application was Python deciding to silently cast NaN falues to
0L, as discussed here:
Christian Heimes [EMAIL PROTECTED] added the comment:
The issue with nan - 0L was fixed in Python 2.6. I can't recall if I
fixed it or somebody else but I know it was discussed.
$ python2.6
long(float(nan))
Traceback (most recent call last):
File stdin, line 1, in module
ValueError: cannot
Raymond Hettinger [EMAIL PROTECTED] added the comment:
Mark, thanks for checking that all the tests still pass. Please do add
back the missing lines in PyObject_RichCompareBool(). It seems that
their removal was not a necessary part of eliminating default sort-ordering.
Mark Dickinson [EMAIL PROTECTED] added the comment:
Taking it out probably had something to do with NaNs, but this
discussion needs to avoid getting lost in NaN paradoxes and instead
focus on a notion of membership that is ALWAYS true given object
identity. This is an essential pragmatism
Christian Heimes [EMAIL PROTECTED] added the comment:
Here is a tes case for 3.0
--
keywords: +patch
Added file: http://bugs.python.org/file11984/issue4296_test.patch
___
Python tracker [EMAIL PROTECTED]
http://bugs.python.org/issue4296
Mark Dickinson [EMAIL PROTECTED] added the comment:
One more data point: in both 2.x and 3.x, sets behave like the 2.x
lists:
Python 3.0rc2+ (py3k:67177M, Nov 10 2008, 16:06:34)
[GCC 4.0.1 (Apple Inc. build 5488)] on darwin
Type help, copyright, credits or license for more information.
s =
Raymond Hettinger [EMAIL PROTECTED] added the comment:
I'm not happy with the 3.0 change. IMO, the best behavior is the one
that allows code reviewers to correctly reason about the code by
assuming basic invariants. In 2.6, all of the following are always true:
assert a in [a]
assert a
Mark Dickinson [EMAIL PROTECTED] added the comment:
Here's a patch incorporating Christian's tests, the missing lines from
PyObject_RichCompareBool, and some additional tests to check that
[x] == [x] and the like remain True even when x is a NaN. Oh, and a
Misc/NEWS entry.
With this patch,
Christian Heimes [EMAIL PROTECTED] added the comment:
Oh h... Raymond, you are so right! I was too tired to think of all
implications related to the different semantic in 3.0, especially the
for in / is in invariant. I'm including Guido and Barry into the
discussion. This topic needs some
Christian Heimes [EMAIL PROTECTED] added the comment:
The tests are passing on Ubuntu Linux 64bit, too.
Good work everybody!
--
assignee: gvanrossum - barry
stage: test needed - patch review
___
Python tracker [EMAIL PROTECTED]
Mark Dickinson [EMAIL PROTECTED] added the comment:
With the lines Raymond mentions restored, all tests (including the extra
tests Christian's posted here) pass for me, and I also get:
a = [float('nan')]
a == a
True
Incidentally, it looks as though the PyObject_RichCompareBool lines were
Raymond Hettinger [EMAIL PROTECTED] added the comment:
Mark, the patch looks good. Thanks. Before committing, please add one
other test that verifies the relationship between in in membership
tests and in in a for-loop:
for constructor in list, tuple, dict.fromkeys, set, collections.deque:
New submission from Michael B Curtis [EMAIL PROTECTED]:
Found in Python 2.4; not sure what other versions may be affected.
I noticed a contradiction with regards to equivalence when experimenting
with NaN, which has the special property that it is itself, but it
doesn't == itself:
a =
Christian Heimes [EMAIL PROTECTED] added the comment:
Interesting, Python 3.0 behaves differently than Python 2.x. Nice catch! :)
Python 3.0rc2 (r30rc2:67177, Nov 10 2008, 12:12:09)
[GCC 4.2.4 (Ubuntu 4.2.4-1ubuntu3)] on linux2
Type help, copyright, credits or license for more information.
nan
24 matches
Mail list logo
