vbvictor wrote: > > I feel this should be part of [lifetime safety > > analysis](https://clang.llvm.org/docs/LifetimeSafety.html), but it > > currently don't model lambda captures/threads I think? > > We do model lambda captures: https://godbolt.org/z/9Y7PYcfsf, but yes, > threads we do not model yet. Though it definitely would be nice to have! (But > I don't think we have any open issues regarding that)
Oh, thank you for clarification! As I see it doesn't support "deeper" indirections when using `std::vector<std::function>`, like this: https://godbolt.org/z/Kvr9xKdcE. Single escapes through single variable works just fine. With that said, I think modeling of `std::thread` will solve half of the logic of this check. The other half could be addressed by modeling "`std::vector<Fun>`" escape paths. https://github.com/llvm/llvm-project/pull/203757 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
