@beckermr, the v0/v1 testing code is becoming horrible. Would you mind me if I tried converting all v0 tests to use the same tempdir approach as v1 — i.e. passed top-level tempdir rather than the recipe subdir?

I imagine it's going to be a lot of work, but it should make everything cleaner. Perhaps going in the other direction (i.e. making v1 tests use recipe subdir) would be less work, but I find the resulting backreferences kinda iffy.

If you do that, could you also migrate to tmp_path? tmpdir returns a py.path.local object, which really is outdated Python.

Codecov Report

Attention: Patch coverage is 90.00000% with 1 line in your changes missing coverage. Please review.

Project coverage is 75.85%. Comparing base (73b9591) to head (3ec10c9).

❌ Your project check has failed because the head coverage (94.03%) is below the target coverage (95.00%). You can increase the head coverage or adjust the target coverage.

@@            Coverage Diff             @@
##             main    #3765      +/-   ##
==========================================
- Coverage   77.34%   75.85%   -1.49%     
==========================================
  Files         134      134              
  Lines       14851    14855       +4     
==========================================
- Hits        11487    11269     -218     
- Misses       3364     3586     +222     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

If you do that, could you also migrate to tmp_path? tmpdir returns a py.path.local object, which really is outdated Python.

Sure can do. Does that mean it's fine for me to proceed, or should I wait for more ACKs?

Yes the testing code is fiddly in general, so improvements are welcome.

Ah, sorry, removed it while debugging the tests, and forgot to put it back. Which means now I'm going to have to debug why the tests pass…

Heh, turned out we weren't checking schema version in MiniMigrator at all. Added a quick fix, now gotta run, will check if this didn't break anything later.

Shall we merge this one @mgorny?

Yes, please.

That said, tests/test_qt_qt_main.py seems to be suffering from network problems now.

It would probably be also worth looking into mocking some network interactions in the test suite too.

I added network retries to that one to ensure it passes. We can up those maybe. I do not think that one can be mocked.

I added network retries to that one to ensure it passes. We can up those maybe. I do not think that one can be mocked.

Right. I suppose we could try mocking the whole logic that grabs the hash, but perhaps that'd go too far.

I am so sorry! I squashed which will make for a bad conflict on your other PR.

🤦