Feature: move tests to nose2#412
Conversation
If the teardown is missing, other tests fail when running all tests.
The tempdir context manager already exists and just not used in this test cases.
|
Ok, looks straightforward, nice work. Do you think we can make appveyor work with this as well? It would be fine to reduce the Pythons to just 1 version, and use something more recent like 3.11 or 3.12 This is what you'd need to edit: |
Ok. Thanks, missed this file. Will look into it and change it. |
@mottosso I updated appvoyer but added python version 3.8 to 3.14. The free threaded python versions are not added but work as I'm tested them locally. |
|
Looks like some builds pick up an older Python, as if And how come there are only 178 tests running, compared to 193 from before? |
c5e8e12 to
055b7a7
Compare
cd52a33 to
e60434a
Compare
e60434a to
151a858
Compare
The image includes python-2.7 by default which causes some errors when running tests. This fix removes any python version in the Path env before adding the required python version for the test.
151a858 to
3240e06
Compare
The previous approach doesn't use the pathlib.Path object as it would be used by a consumer of the framework. Therefore the tests are always failing on Linux even if the path is correctly registered.
|
It's starting to smell like vibe coding and I'm cautious about spending much time to review this. |
Hello @mottosso, I'm not using any LLM or the like. I'm not a fan of them either. I just try to get the test to run on both Linux and Windows without changing them. As I don't have a Windows PC, I have to rely on the CI to see if the change does not introduce a regression. I did change the commit in place and force pushed for testing purpose only. After the testing I did do a rebase to clean up the chaotic commits. The missing tests from last time where the doc tests. I didn't configure them correctly in nose2 config. If there is a way to have a complete clean environment for each time, please point me to the documentation, I could not find any relevant reference. Out of curiosity, what did give the impression that this is vibe coded? Kind regards, |
|
Ok, I think this looks good and proper. There are a lot of edits, which makes any PR hard to review. But the edits appear mostly identical across many files and I can't spot anything off. Great work. |
Hello,
as with the comment in #409 (comment) by @mottosso this PR switches the tests from nose to nose2.
Kind regards,
Wilfried