hazohelet added a comment. In D156604#4549524 <https://reviews.llvm.org/D156604#4549524>, @tbaeder wrote:
> @hazohelet I changed this to just return `CallExpr->getSourceRange()` and to > not save a source range separately. Can you give this a look please? I was thinking about removing `CallLoc` some time ago, and concluded that we shouldn't. Destructor calls are usually not explicitly written by the user, so `CallExpr` is set to null. `CallLoc` is often set to the declaration location (https://github.com/llvm/llvm-project/blob/97cddb78502eee583b5f4ee02c59b7156398587f/clang/lib/AST/ExprConstant.cpp#L703-L708). So we'll lose this information if we are to remove `CallLoc` from the stack frame, and it seems to be the reason for the test failures. About getting source range from `CallExpr->getSourceRange`, I don't see any obvious problems. All explicitly-written function calls should have its AST in `CallExpr` field. So using the `CallLoc` for the diagnostics location and `CallExpr->getSourceRange` for the source range would do the trick. CHANGES SINCE LAST ACTION https://reviews.llvm.org/D156604/new/ https://reviews.llvm.org/D156604 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
