Fixing byte-order consistency/mismatch for cross-endian platform

[pradghos]

• Addressing dtype.byteorder is not consistent during cross platform joblib.load()  #1123

[pradghos]

Circle CI failed at ./continuous_integration/build_doc.sh

result = self._call_chain(self.handle_open, protocol, protocol +
File "/home/circleci/miniconda/envs/testenv/lib/python3.9/urllib/request.py", line 494, in _call_chain
  result = func(*args)
File "/home/circleci/miniconda/envs/testenv/lib/python3.9/urllib/request.py", line 1389, in https_open
  return self.do_open(http.client.HTTPSConnection, req,
File "/home/circleci/miniconda/envs/testenv/lib/python3.9/urllib/request.py", line 1349, in do_open
  raise URLError(err)
urllib.error.URLError: <urlopen error [Errno 110] Connection timed out>

Makefile:11: recipe for target 'html' failed

Before this commit https://github.com/joblib/joblib/pull/1181/commits/d9e5adbcc5478159310ebd58b3c070c5feea7e7d all were green except ci/circleci: build

Now joblib.joblib (Testing linux_py37_distributed) failed at test_nested_loop_error_in_grandchild_resource_tracker_silent which looks like unrelated to this PR.

Any pointers would really help . Thank you !

[pradghos]

@ogrisel It would be great if you can take a look. Thanks in advance !

[lesteve]

Thanks a lot for this PR! What would be great is to have a test for this.

Is there a way to store a pickle string obtained from a big endian machine and test that it can be loaded fine in the CI (little endian machine)?

I think we can ignore the CI failures as they do seem spurious.

[ogrisel]

Alternatively could you just please add a couple of unit tests for _is_numpy_array_byte_order_mismatch alone (independently of jolib.load / dump)?

[codecov]

Codecov Report

Merging #1181 (f8a2f16) into master (754433f) will increase coverage by 0.02%.
The diff coverage is 94.73%.

@@            Coverage Diff             @@
##           master    #1181      +/-   ##
==========================================
+ Coverage   94.51%   94.54%   +0.02%     
==========================================
  Files          47       47              
  Lines        6997     7034      +37     
==========================================
+ Hits         6613     6650      +37     
  Misses        384      384              

Impacted Files	Coverage Δ
joblib/test/test_numpy_pickle.py	93.33% <92.59%> (-0.04%)	⬇️
joblib/numpy_pickle.py	98.01% <100.00%> (+<0.01%)	⬆️
joblib/numpy_pickle_compat.py	91.08% <100.00%> (+0.18%)	⬆️
joblib/numpy_pickle_utils.py	93.25% <100.00%> (+0.57%)	⬆️
joblib/test/test_store_backends.py	97.14% <0.00%> (+5.71%)	⬆️

---

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 754433f...f8a2f16. Read the comment docs.

[pradghos]

@lesteve @ogrisel Thank you very much for the review. I have added test. Please let me know in case of any further comments.

[ogrisel]

Some nitpick. I plan to push a new commit to better test the byteswap operation itself after that.

[ogrisel]

Merged!, thanks @pradghos!