Send one character (no newline) to stdout in protocol test to guarantee a single "message" and thus a single custom value

[christian-monch]

Fixes issue #4921

This PR modifies the test datalad.support.tests.test_witless_runner.test_parameterized_protocol()
to only send a single character to stdout of the subprocess, i.e. 1. Previously two characters were
sent, i.e. 1 and \n. Depending on the behavior of the stdout-pipe, the data might be split up or
delivered in toto. In the first case, which is rare, the protocol.pipe_data_received() will be called
twice, leading to an unexpected result.

Changelog

🐛 Bug Fixes

• Fix a faulty test that made wrong assumptions about pipe-behavior. Fixes 0.13.3 on conda and fresh git-annex: duplicated output picked up within test_runner_parametrized_protocol? #4921

[qlty-cloud-legacy]

Code Climate has analyzed commit bc4ded8 and detected 0 issues on this pull request.

View more on Code Climate.

[yarikoptic]

• ideally should be against maint not master
• moving (will remove there) my question/comment from the underlying issue:

Thanks @christian-monch for the analysis!!! I fail to see how/why changing the test "fixes" the situation in christian-monch@bc4ded8#diff-472642261afcc22bd6f543b4a8c20533588ec9281bd964137b29f79885e56164L197 ? Test code seems to be totally legit - just running a command which prints a single character, so should not be changed. IMHO it is the code of the runner is wrong somewhere that it manages to duplicate that single character, so better be fixed in the code (not in the test). Or what am I missing?

[codecov]

Codecov Report

Merging #6978 (bc4ded8) into maint (29b42b4) will increase coverage by 0.54%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##            maint    #6978      +/-   ##
==========================================
+ Coverage   90.13%   90.68%   +0.54%     
==========================================
  Files         354      354              
  Lines       46326    47355    +1029     
  Branches     6613     6609       -4     
==========================================
+ Hits        41755    42942    +1187     
+ Misses       4554     4397     -157     
+ Partials       17       16       -1     

Impacted Files	Coverage Δ
datalad/runner/tests/test_witless_runner.py	95.39% <100.00%> (+2.40%)	⬆️
datalad/support/tests/test_annexrepo.py	97.97% <100.00%> (ø)
datalad/distribution/create_sibling.py	52.95% <0.00%> (-16.85%)	⬇️
datalad/local/remove.py	88.73% <0.00%> (-8.29%)	⬇️
datalad/support/tests/test_repo_save.py	95.34% <0.00%> (-4.66%)	⬇️
datalad/support/gitrepo.py	89.56% <0.00%> (-3.19%)	⬇️
datalad/tests/utils_pytest.py	87.05% <0.00%> (-3.08%)	⬇️
datalad/dataset/gitrepo.py	96.15% <0.00%> (-0.67%)	⬇️
datalad/runner/gitrunner.py	92.19% <0.00%> (-0.31%)	⬇️
datalad/_version.py	45.68% <0.00%> (-0.28%)	⬇️
... and 5 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[christian-monch]

• ideally should be against maint not master
• moving (will remove there) my question/comment from the underlying issue:

Thanks @christian-monch for the analysis!!! I fail to see how/why changing the test "fixes" the situation in christian-monch@bc4ded8#diff-472642261afcc22bd6f543b4a8c20533588ec9281bd964137b29f79885e56164L197 ? Test code seems to be totally legit - just running a command which prints a single character, so should not be changed. IMHO it is the code of the runner is wrong somewhere that it manages to duplicate that single character, so better be fixed in the code (not in the test). Or what am I missing?

The test code did NOT send a single character! It sent two characters. The reason is that the print-function appends an "\n" to the string that it writes out. Unless you specify the keyword-argument end to be something else, in this case we specify end="". Then we get only a single character printed.

❱ python -c "print('1')"|wc -c                                                                                                                                                                               130 !
2
❱ python -c "print('1', end='')"|wc -c
1

[christian-monch]

• ideally should be against maint not master

Thx. Fixed!
I actually branched of maint. Always forget to change the merge-target in the PRs ;-)

[yarikoptic]

The test code did NOT send a single character! It sent two characters

you kept saying that and I was "why he keeps saying that?" ;) Looked at the test code and the purpose of that test and got it - indeed there is an assumption in the test that there would be a single message/invocation of the pipe_data_received where we replace with our custom (5) value. And that made me understand @christian-monch ! ;-) Thanks! Let's proceed - I can't resist such a green CI ;-)

[github-actions]

🚀 PR was released in 0.17.4 🚀