Repository navigation
Update VT_ARRAY | VT_RECORD tests. - #982
Merged
Merged
Conversation
- Define `_SafeArrayGetRecordInfo` at module level using c_void_p in argtypes instead of `POINTER(IRecordInfo)`, eliminating the need for local import. - Use byref for result parameter following surrounding code style. - Add docstring and type ignore comment for return value.
Since commit 0681ffe, unpacking VT_ARRAYs containing VT_RECORDs no longer raises `KeyError('36')`. Update the characterization tests for `WrapBstrRecord`, `WrapValueRecord`, and `ReverseValueRecords` to verify the expected return values on the normal path instead of expecting a `KeyError`.
This test uses `faulthandler` and `multiprocessing` to run the test in a separate process.
The test `call_reverse_bstr_records` was an existing test. This change converts it to a characterization test. It captures and verifies the crash without breaking the whole test suite. The test probably crashes because of a double-free bug. Root cause (hypothesis): - The `VARIANT` has a `SAFEARRAY`. This `SAFEARRAY` probably shares the same `BSTR` pointer as `lp`. - After the COM call returns, `VariantClear` runs. `VariantClear` probably frees that `BSTR`. - Then Python runs `del lp`. This probably tries to free the same `BSTR` again. - This probably causes a double-free and heap corruption.
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## up1_followup_104 #982 +/- ##
===================================================
Coverage ? 90.72%
===================================================
Files ? 147
Lines ? 15344
Branches ? 0
===================================================
Hits ? 13921
Misses ? 1423
Partials ? 0 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
Collaborator
Author
|
While I was working on this, Python 3.15 crashed. So I wrote a test to capture the crash. I was surprised by the CI result. This means #104 may have worked as it was, if it had been merged back then. This is a delicate memory issue. I think some changes inside Python are related, but I don't fully understand it yet. Anyway, in the next PR, I will make the production code work as intended on every supported Python version. |
Updated to handle different Python versions.
Since `LP_SAFEARRAY` is also `_Pointer[tagSAFEARRAY]`, `_Pointer[tagSAFEARRAY] | LP_SAFEARRAY` was redundant.
junkmd
force-pushed
the
record_safearray_characterization_tests
branch
from
November 4, 2026 13:09
83d03d0 to
893f402
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Related to #63
The fix itself will be done in a separate PR.