-
-
Notifications
You must be signed in to change notification settings - Fork 10.9k
PyPy is crashing on some tests. #17420
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
Comments
I should run do a full leak check run on master some time soon, there are definitely some ref-counting issues introduced in all the recent code changes. I have tried running the first 2% (both master and that PR) of the test suit on python-debug with leak checking, but while there are some ref-leaks, there is no change, but I guess that doesn't mean there can't be a refcounting issue, but it makes it less likely? |
Thanks!
At least less likely that a refcounting issue causes the testing failure :) The changes are mostly in error handling code and I suspect much of that is not tested. I should probably set up to test with PyPy even though it doesn't support recent Python. One advantage a lot of commits in a PR is that bisecting is easy. OTOH, this may have exposed a problem in PyPy. The test seems to pull from the head of the development branch. |
Note the linux PyPy passes. It might be an issue with |
To put it somewhere, and because its not completely useless maybe to see, here is the output of a full leak-check run. Note that a few of these are expected. The f2py leaks were always there IIRC, the array-coercion (and some others) are due to the buffer "cache", some are known leaks (and marked, but I did not skip it). If the Some that are pretty certainly real: The random ones (likely all the same issue?), the EDIT: Hmm, I think quite a few of these may be just be issues with pytest-leaks in combination to some fixtures or maybe some other loaded pytest plugin...
|
Segfault on Linux. Another in the same place here.
|
This seems to be fixed. |
See here. @mattip How does one discover what the problem is with PyPy. My guess is that it is some reference problem, as PyPy doesn't do reference counting and there is a lot of it in this PR.
Reproducing code example:
Error message:
NumPy/Python version information:
The text was updated successfully, but these errors were encountered: