The pypy builds of pywin32 do not contain anything useful

[nehaljwani]

$ tar -tvf pywin32-304-py38h32ec214_2.tar.bz2
-rw-rw-rw-  0 0      0          25 Oct 28 20:11 info/test/test_time_dependencies.json
-rw-rw-rw-  0 0      0          42 Oct 28 20:11 info/paths.json
-rw-rw-rw-  0 0      0         109 Oct 28 20:11 info/hash_input.json
-rw-rw-rw-  0 0      0         332 Oct 28 20:11 info/index.json
-rw-rw-rw-  0 0      0         535 Oct 28 20:05 info/recipe/0003-remove-PATH-pth-hack.patch
-rw-rw-rw-  0 0      0         643 Oct 28 20:11 info/recipe/conda_build_config.yaml
-rw-rw-rw-  0 0      0         727 Oct 28 20:05 info/recipe/do-not-build-scintilla.patch
-rw-rw-rw-  0 0      0        1520 May  2 08:48 info/licenses/Pythonwin/License.txt
-rw-rw-rw-  0 0      0        1568 Oct 28 20:05 info/recipe/recipe-scripts-license.txt
-rw-rw-rw-  0 0      0        1733 Oct 28 20:11 info/recipe/meta.yaml
-rw-rw-rw-  0 0      0        1831 Oct 28 20:05 info/recipe/run_test.py
-rw-rw-rw-  0 0      0        2115 Oct 28 20:11 info/test/run_test.py
-rw-rw-rw-  0 0      0        3356 Oct 28 20:05 info/recipe/meta.yaml.template
-rw-rw-rw-  0 0      0        7300 Oct 28 20:11 info/about.json
-rw-rw-rw-  0 0      0           0 Oct 28 20:11 info/git
-rw-rw-rw-  0 0      0           0 Oct 28 20:11 info/files

[nehaljwani]

If this is expected, how do we make the pip check pass? Should we make it a dummy module as well? cc @mariusvniekerk

[hmaarrfk]

I'm not sure it is expected, my guess is that it was done by mistake.

Selectors are used on every line of substance, while a skip: true # [win pypy] should have liked been used. Portalocker (mentionned above) hit this issue.

[akrherz]

My googling for a status on this issue leads me to missing upstream support here: mhammond/pywin32#1289

and a heroic PR attempt here: #41