@microsoft-github-policy-service agree

Hi, is there any update here? I can't get debugging to work on macOS using PySide6 and Matplotlib, with all mentioned tricks. Or is there any other hint I am overseeing?

@wmvanvliet can you close and reopen? Not sure why the lint portion is waiting but I don't think it's going to run since it's been waiting for 4 months.

The linting is not running because it requires approval from a maintainer.

/AzurePipelines run

It's running now :) We have to manually run it with /azp run when it comes from a fork, for security reasons.

@AdamYoblick does this test failure look like the flakey test you added retry to? Maybe we need to merge this branch with main?

/AzurePipelines run

@AdamYoblick does this test failure look like the flakey test you added retry to? Maybe we need to merge this branch with main?

Yes, it does look similar. Test workers crashing. Although I don't think merging main in will make a difference in this case because the test I marked as flaky was not the test method that failed here (test_log_dir_env). Unfortunately, we probably need to mark more tests as flaky. I don't remember 100% but I think that it was only python 3.12 on windows crashing before, just like in this case.

I wonder if we just need to mark the entire test suite as flaky with retries until this is properly investigated.

This one is a pytest crash reading from a socket, so likely to be environment-induced.

In general I recall two main categories of flaky tests. One was anything to do with attach-to-PID, and specific to tests that do that. The other is a general failure when sockets just get randomly closed or timed out, which I have only ever seen in CI and not locally, but can happen with almost every test (since anything that runs debugpy involves sockets); I always suspected it is also something coming from the environment for those runs but I never saw any clues as to what it could be.

Thank you all for taking care, may I please ask if approved now (If I understand correctly?), when it can be released? Or how can I help to test e.g. in my environment?

@SebusFu - You don't have to wait for a debugpy release to be able to test the changes. Here's what you do:

1. Clone or fork the repo (which you already have)
2. Add the following line to your launch configuration in launch.json:

"debugAdapterPath": "<path_to_your_fork>/src/debugpy/adapter"

3. Launch the file

This will use your local debugpy source instead of the debugpy that comes installed with the Python Core Extension, and you can see if it works.

Thanks!

Thank you very much, works like a charm now for me!

/azp run

This test fail has nothing to do with your changes, it's a flaky test, so I'm going to merge this in.