[issue26991] Possible reference leak in MAKE_FUNCTION

Roundup Robot added the comment: New changeset eaa3a71a6f62 by Benjamin Peterson in branch '3.5': fix possible refleak in MAKE_FUNCTION (closes #26991) https://hg.python.org/cpython/rev/eaa3a71a6f62 New changeset 7270701cf5bc by Benjamin Peterson in branch 'default': merge 3.5 (#26991) https://h

[issue26991] Possible reference leak in MAKE_FUNCTION

Serhiy Storchaka added the comment: I don't know what is the best place for this function. You can just break compiling functions with annotations, run tests, found tests that use functions with annotations, and add new test near one of them. You can test your test by removing the fix and runn

[issue26991] Possible reference leak in MAKE_FUNCTION

Xiang Zhang added the comment: I'm willing to. But I don't know where is the suitable place to put such a test and since the reference leak only appears in error case, can compile a function with annotations in a loop accomplish the test goal? -- __

[issue26991] Possible reference leak in MAKE_FUNCTION

Serhiy Storchaka added the comment: LGTM. Would be nice to have a test (just compile a function with annotations in a loop). -- components: +Interpreter Core stage: -> test needed type: -> behavior ___ Python tracker

[issue26991] Possible reference leak in MAKE_FUNCTION

Xiang Zhang added the comment: Sorry, wrong patch. Attach the right one. -- Added file: http://bugs.python.org/file42795/make_function_reference_leak.patch ___ Python tracker ___

[issue26991] Possible reference leak in MAKE_FUNCTION

New submission from Xiang Zhang: *names* is not Py_DECREFed in MAKE_FUNCTION in error case. -- files: make_function_reference_leak.patch keywords: patch messages: 265235 nosy: serhiy.storchaka, xiang.zhang priority: normal severity: normal status: open title: Possible reference leak in M