Skip test_source_mtime_long_long on 32bit and lower platforms

[mimi1vx]

• Create a new changelog file in the changelog folder, with a name like <ISSUE NUMBER>.<TYPE>.rst. See changelog/README.rst for details.
• Target the master branch for bug fixes, documentation updates and trivial changes.
• Target the features branch for new features and removals/deprecations.
• Include documentation when adding new features.
• Include new tests or update existing tests when applicable.

Unless your change is trivial or a small documentation fix (e.g., a typo or reword of a small section) please:

• [x ] Add yourself to AUTHORS in alphabetical order;

[codecov]

Codecov Report

Merging #5045 into master will decrease coverage by 0.06%.
The diff coverage is 100%.

@@            Coverage Diff            @@
##           master   #5045      +/-   ##
=========================================
- Coverage   96.06%     96%   -0.07%     
=========================================
  Files         114     114              
  Lines       25768   25769       +1     
  Branches     2548    2548              
=========================================
- Hits        24754   24739      -15     
- Misses        704     715      +11     
- Partials      310     315       +5

Impacted Files	Coverage Δ
testing/test_assertrewrite.py	84.06% <100%> (+0.02%)	⬆️
src/_pytest/assertion/util.py	93.38% <0%> (-4.29%)	⬇️
src/_pytest/unittest.py	93.12% <0%> (-1.06%)	⬇️
src/_pytest/doctest.py	95.69% <0%> (-0.72%)	⬇️
testing/test_pdb.py	98.99% <0%> (-0.21%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 973301b...77526f4. Read the comment docs.

[blueyed]

Thanks!

The test was added for #4903 - /cc @bmwiedemann

Just for future reference: you should use "Fixes …" in commit message, especially given that you have created an issue already.

[bmwiedemann]

hmm. Wasnt there this size & 0xffffffff that is supposed to handle overly large files? If something is broken there, there should be a better fix than this.

waiting for feedback

[mimi1vx]

@bmwiedemann python Integer bigger that 2^31-1 can't be converted to 'int' on a platform where sys.maxsize is 2^31-1 .. so this test is wrong on this platforms, and proprely cast OverflowError Exception

[bmwiedemann]

The test is not doing a cast to int. And the code specifically uses "L" for unsigned to have a 0..2^32-1 range. Maybe including a backtrace would make the problem clearer.

[nicoddemus]

The traceback was included in #5046 (comment):

[  348s]         timestamp = 2 ** 32 + offset
[  348s] >       os.utime(str(p), (timestamp, timestamp))
[  348s] E       OverflowError: Python int too large to convert to C long

And seems to be failing for both -1 and +1 offsets.

Strangely, this passes for me on 32-bit Python 2.7 on Windows:

λ python -c "import platform;print platform.architecture()"
('32bit', 'WindowsPE')

λ pytest testing\ -k test_source_mtime_long_long -v
======================== test session starts ========================
platform win32 -- Python 2.7.13, pytest-4.4.1.dev51+g1ed4ae863, py-1.6.0, pluggy-0.9.0
...

testing/test_assertrewrite.py::test_source_mtime_long_long[-1] PASSED [ 50%]
testing/test_assertrewrite.py::test_source_mtime_long_long[1] PASSED [100%]

============= 2 passed, 2383 deselected in 3.30 seconds =============

On 64-bits Python I get, as expected:

λ py -3.6 -c "import platform;print(platform.architecture())"
('64bit', 'WindowsPE')

I would expect to be able to reproduce this...

[nicoddemus]

@mimi1vx any updates on this? Can you please post the output of:

λ python -c "import platform;print platform.architecture()"
('32bit', 'WindowsPE')

In your system?

[mimi1vx]

@nicoddemus

[    3s] + python2 -c 'import platform;print platform.architecture()'
[    3s] ('32bit', 'ELF')

[nicoddemus]

@mimi1vx interesting, thanks. Which OS is that?

[mimi1vx]

@nicoddemus ELF - > linux

btw>

abuild@penguin:~> python2 
Python 2.7.16 (default, Mar 04 2019, 07:13:50) [GCC] on linux2
>>> timestamp = 2 ** 32 +1
>>> import os
>>> os.utime(str("/etc/os-release"), (timestamp, timestamp))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
OverflowError: Python int too large to convert to C long

abuild@penguin:~> python3
Python 3.7.2 (default, Dec 30 2018, 16:18:15) [GCC] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import os
>>> timestamp = 2 ** 32 +1
>>> os.utime(str("/etc/os-release"), (timestamp, timestamp))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
OverflowError: timestamp out of range for platform time_t
>>> 

[nicoddemus]

I see, thanks!

Well now that's been clarified, I think it is harmless to merge this, otherwise the test blows up anyway. 👍