-
-
Notifications
You must be signed in to change notification settings - Fork 79
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
6.9.0: DiscoveryTests.test_package_discovery
fails
#479
Comments
As discovered in #463, the issue appears to be that something in the Python environment is causing modules to be reloaded and thus the object imported for Today I learned that OpenIndiana is a FreeBSD derivative, so I'm not able to readily replicate any issue. I'll either need a donor environment in which I can test or I'll need someone with OpenIndiana to inspect the environment and identify the failed expectation in importlib_metadata.
Did the issue emerge with 6.9.0; did tests pass on earlier versions? |
Sorry, no. OpenIndiana is Solaris descendant. Actually, it it an OpenSolaris fork. It has almost nothing to do with FreeBSD.
I can try to look at it, but I'd need some suggestions where to start.
Short answer: I do not know. Long answer follows :-). A bit of history: I filled this issue during my work on upgrading the OpenIndiana importlib_metadata package from version 6.6.0 to version 6.9.0. With version 6.6.0 we supported both Python 3.7 and Python 3.9. When I ran test for version 6.6.0 I noticed that all tests passed for Python 3.7, while some tests failed for Python 3.9. IIRC there were 5 to 10 such failing tests. I leaved this as is since we had at least one flavor (for Python 3.7) passing all tests. With the 6.7.0 release IIRC some test(s) started to fail for Python 3.7 too, so I decided to stay at 6.6.0 until I find some time to analyze issues and possibly report bugs here (testing with Python 3.9 was still failing in more than few tests). With EOL of Python 3.7 in June we no longer planned to build importlib_metadata for it. So passing (or almost passing) testing for Python 3.7 was no longer enough. Version 6.8.0 was failing with Python 3.9 too, but I do not remember how many tests failed. I just skipped this release. If needed, I can try to test. When version 6.9.0 was released recently I finally looked closely and I found that only 4 tests are failing here with Python 3.9. This one reported here, two reported in #463, and one more that expects there is no |
Please note that although we run tests for So I tried to run tests in fully isolated environment using:
and all of them passed. So yes, something in the real environment is affecting tests. |
I just tried to test
List of installed Python packages:
|
This behavior reinforces my my suspicion that something in the environment is preventing |
My suggestion for a next step - in the environment where the tests are failing, add a breakpoint here in the install routine and inspect what's happening in |
I'm not longer able to reproduce this issue. Tested with versions |
I'm running tests for
importlib_metadata 6.9.0
and I noticed that the following test fails:Tested on OpenIndiana with Python 3.9.16.
The text was updated successfully, but these errors were encountered: