Odd behavior when using datetime.timedelta under cProfile

[beaugunderson]

BPO	35004
Nosy	@asottile, @pganssle, @tirkarthi, @beaugunderson, @isidentical

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = None
created_at = <Date 2018-10-16.22:31:03.052>
labels = ['3.7']
title = 'Odd behavior when using datetime.timedelta under cProfile'
updated_at = <Date 2019-11-18.19:14:19.824>
user = 'https://github.com/beaugunderson'

bugs.python.org fields:

activity = <Date 2019-11-18.19:14:19.824>
actor = 'BTaskaya'
assignee = 'none'
closed = False
closed_date = None
closer = None
components = []
creation = <Date 2018-10-16.22:31:03.052>
creator = 'beaugunderson'
dependencies = []
files = []
hgrepos = []
issue_num = 35004
keywords = []
message_count = 4.0
messages = ['327846', '327856', '328268', '356901']
nosy_count = 5.0
nosy_names = ['Anthony Sottile', 'p-ganssle', 'xtreak', 'beaugunderson', 'BTaskaya']
pr_nums = []
priority = 'normal'
resolution = None
stage = None
status = 'open'
superseder = None
type = None
url = 'https://bugs.python.org/issue35004'
versions = ['Python 3.7']

[beaugunderson]

In chasing down a bug in my code that only manifests itself when running under cProfile I managed to find the culprit in datetime.timedelta by way of dateutil.

Here is a small repro:

https://gist.github.com/beaugunderson/68ea6424a4fdcf763ccad08e42a74974

I believe an exception should still be raised in broken() when run under cProfile, but for some reason it is not.

[asottile]

Here's a simpler reproduction without involving a third party library:

>>> import cProfile
>>> from datetime import timedelta
>>> pr = cProfile.Profile()
>>> timedelta.total_seconds(-25200)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: descriptor 'total_seconds' requires a 'datetime.timedelta' object but received a 'int'
>>> pr.enable()
>>> timedelta.total_seconds(-25200)
2177366085.870893
>>> pr.disable()
>>> timedelta.total_seconds(-25200)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: descriptor 'total_seconds' requires a 'datetime.timedelta' object but received a 'int'

However, it appears fixed on master:

$ ./python
Python 3.8.0a0 (heads/master:c984d20ec8, Oct 16 2018, 20:47:49) 
[GCC 7.3.0] on linux
>>> 
>>> import cProfile
>>> from datetime import timedelta
>>> pr = cProfile.Profile()
>>> timedelta.total_seconds(-25200)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: descriptor 'total_seconds' requires a 'datetime.timedelta' object but received a 'int'
>>> pr.enable()
>>> timedelta.total_seconds(-25200)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: descriptor 'total_seconds' for 'datetime.timedelta' objects doesn't apply to 'int' object
>>> pr.disable()
>>> timedelta.total_seconds(-25200)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: descriptor 'total_seconds' requires a 'datetime.timedelta' object but received a 'int'

Doing a git bisect shows that this commit fixed the issue: #8300

Here's the bpo for that: https://bugs.python.org/issue34126

[tirkarthi]

Thanks Anthony for the details. This was fixed in master and back ported to 3.7. The fix is also released with 3.7.1 [0] which I have verified locally. So I propose closing this issue since upgrading to the latest maintenance release fixes this.

[0] https://www.python.org/downloads/release/python-371/

Thanks

[isidentical]

This bug is fixed in both 3.8 and 3.7, @p-ganssle what do you think about closing this?

[StanFromIreland]

This bug has been fixed for a few years, this issue can be closed.