https://github.com/NagyDonat commented:
> Im sorry about messing up the naming convention, I havent contributed too > much to open source stuff. No harm done, this is part of the plan :slightly_smiling_face: Every contributor must learn these sometime and this time it was your turn to do so. ----- I read the suggested code changes and overall I like them, this seems to be a useful change and the code seems to be nice. However, as I started the CI processes, it seems that several tests fail with the current revision of the commit. You can find the details of the failures e.g. by searching for the string "FAIL: Clang" in the (very long...) [raw command line output](https://productionresultssa19.blob.core.windows.net/actions-results/fc316f15-7917-4184-97b0-6a08fac47e2b/workflow-job-run-2dcc4368-b2ed-502a-9023-83976bbc2898/logs/job/job-logs.txt?rsct=text%2Fplain&se=2025-08-27T12%3A38%3A14Z&sig=AQ52BAVVkoiFPBdeWcN3blpDMidXGH4bXOPsZtIHvs0%3D&ske=2025-08-27T23%3A07%3A54Z&skoid=ca7593d4-ee42-46cd-af88-8b886a2f84eb&sks=b&skt=2025-08-27T11%3A07%3A54Z&sktid=398a6654-997b-47e9-b12b-9515b896b4de&skv=2025-05-05&sp=r&spr=https&sr=b&st=2025-08-27T12%3A28%3A09Z&sv=2025-05-05). ----- By the way it seems that the pull request https://github.com/llvm/llvm-project/pull/152751 (which is also under review right now) proposes a very similar, but more constrained heuristic: it also extends the `checkPostCall` callback of MallocChecker, but it only marks pointers passed to a constructor as "escaped" if it is the constructor of either `unique_ptr` or `shared_ptr`. (That commit also introduces another heuristic which checks `unique_ptr` or `shared_ptr` fields of temporary objects to avoid a different sort of false positive.) We'll need to pay attention to avoid introducing redundancies or conflicting heuristics. https://github.com/llvm/llvm-project/pull/155131 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
