BF: if early attempt to parse resulted in error, setup subparsers

[yarikoptic]

Without that using a short option for a command which is also
present on top level results in early attempt error and then no
parsers even tried, thus giving that error from the main/top parser to the user

Fixes #6814

Regression was introduced in 0b4937d ( AKA 0.16.0^268^217 )

I have not investigated possible impact on completion etc, but when tried - still worked fine.

[yarikoptic]

wow this hit some nerve:

FAILED ../datalad/cli/tests/test_main.py::test_script_shims[datalad] - datala...
FAILED ../datalad/cli/tests/test_main.py::test_script_shims[git-annex-remote-datalad-archives]
FAILED ../datalad/cli/tests/test_main.py::test_script_shims[git-annex-remote-datalad]
FAILED ../datalad/cli/tests/test_main.py::test_cfg_override - FileNotFoundErr...
FAILED ../datalad/cli/tests/test_main.py::test_librarymode - datalad.runner.e...
FAILED ../datalad/cli/tests/test_main.py::test_completion - datalad.runner.ex...
FAILED ../datalad/core/distributed/tests/test_clone.py::test_invalid_args - d...
FAILED ../datalad/core/distributed/tests/test_clone.py::test_clone_crcns - Fi...
FAILED ../datalad/core/distributed/tests/test_clone.py::test_clone_datasets_root
FAILED ../datalad/core/distributed/tests/test_clone.py::test_clone_simple_local
FAILED ../datalad/core/distributed/tests/test_clone.py::test_clone_dataset_from_just_source
FAILED ../datalad/core/distributed/tests/test_clone.py::test_clone_isnot_recursive
FAILED ../datalad/core/distributed/tests/test_clone.py::test_clone_into_dataset
FAILED ../datalad/core/distributed/tests/test_clone.py::test_notclone_known_subdataset
FAILED ../datalad/core/distributed/tests/test_clone.py::test_clone_missing_commit
FAILED ../datalad/core/distributed/tests/test_clone.py::test_reckless[True]
FAILED ../datalad/core/distributed/tests/test_clone.py::test_reckless[False]
FAILED ../datalad/core/distributed/tests/test_clone.py::test_install_source_relpath

I guess I or someone else would need to look into it in greater depth :-/

edit : with a sober eye -- all those FAILs simply can't relate to the change since have nothing to do with the parser -- what was going on?

[mih]

I triggered a rerun of the failed appveyor runs to get a sense of whether the widespread errors are a consequence of some service failure.

Re fix: Superficially, this makes sense. Even though I did the last RF of this code, it is hard for me to say "OK" with confidence. The last RF was focused on making things faster, but it certainly did not aim for (or achieve) an improvement of this code's comprehensibility.

[yarikoptic]

bad news: somehow all those FAILED are "real". good news: python -m pytest -s -v -k 'test_main' seems to reproduce one of the failures in test_cfg_override for me locally (but not if I just run -k test_cfg_override -- so smells of some side effect somewhere). I will try to look into it later (on the move in the coming hours)

[yarikoptic]

gy gy - it was the new test I added. I didn't realize that run_main does it in the process and doesn't run external datalad, so that -C I added was chdir'ing test process and causing all this turmoil. I rewrote and force pushed (twice ;-) ). Should be all good/only relevant fails now ;)

[codecov]

Codecov Report

Merging #6980 (9b21010) into maint (3300769) will increase coverage by 0.99%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##            maint    #6980      +/-   ##
==========================================
+ Coverage   90.14%   91.13%   +0.99%     
==========================================
  Files         354      354              
  Lines       46321    46325       +4     
  Branches     6612     6612              
==========================================
+ Hits        41754    42218     +464     
+ Misses       4551     4091     -460     
  Partials       16       16              

Impacted Files	Coverage Δ
datalad/cli/parser.py	86.03% <ø> (+1.80%)	⬆️
datalad/cli/tests/test_main.py	95.34% <100.00%> (+0.11%)	⬆️
datalad/_version.py	45.68% <0.00%> (-0.28%)	⬇️
datalad/cli/interface.py	100.00% <0.00%> (+2.77%)	⬆️
datalad/tests/utils.py	65.10% <0.00%> (+11.03%)	⬆️
datalad/tests/test_tests_utils.py	92.34% <0.00%> (+92.34%)	⬆️

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

[yarikoptic]

the only fail travis is a known #4496 (comment) . As @mih gives some support in the fix ("makes sense") , we have a unittest for it now, and it seems nothing breaks -- I will proceed with merge since unlikely we would get a better fix (if there is one possible) any time soon .

[github-actions]

🚀 PR was released in 0.17.4 🚀