bpo-38962: Fix reference leak in the per-subinterpreter gc#17457
bpo-38962: Fix reference leak in the per-subinterpreter gc#17457miss-islington merged 2 commits into
Conversation
78add69 to
d0de63a
Compare
December 3, 2019 22:14
|
I'm not super convinced about this PR, but indeed it fixes the reference leaks in test__xxsubinterpreters (when all other PRs in https://bugs.python.org/issue38962 are applied first): ❯ ./python -m test test__xxsubinterpreters -R : == Tests result: SUCCESS == 1 test OK. Total duration: 17.6 sec |
Sorry, something went wrong.
d0de63a to
8f036d2
Compare
December 4, 2019 10:56
|
@vstinner I have modified the PR to trigger the collection only on sub interpreters. Can you check it out? |
Sorry, something went wrong.
0e8f8c5 to
4f9a89b
Compare
December 4, 2019 11:29
4f9a89b to
487863c
Compare
December 4, 2019 11:30
vstinner
left a comment
There was a problem hiding this comment.
LGTM.
As I wrote, I'm not fully satisfied by calling _PyGC_CollectNoFail() after PyInterpreterState_Clear() because of the risk of bugs in objects finalizers, but we cannot fix all Python finalization issues at once. Let's unblock the buildbots, and only later investigate for a better design.
Sorry, something went wrong.
https://bugs.python.org/issue38962
Automerge-Triggered-By: @pablogsal