https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84883
--- Comment #2 from Tim Ruehsen <tim.ruehsen at gmx dot de> --- (In reply to Martin Sebor from comment #1) > (b) could be implemented in the C/C++ front ends which do have access > to the algorithm but not to data flow analysis, so the warning there would > be quite simplistic (basically limited to type and name matching) and could > be prone to false positives. If someone makes up a patch and give it to me, I can take the time to test with a decent amount of projects. So that we can get figures about how often the bug triggers and how many false positives there are. Without this, you are guessing. And I meanwhile had two occurrences where such an option would have saved several hours of work.