Sirraide wrote: > I'm not sure if that's reasonable, or if the first warning should be omitted. Hmm, I suspect that not doing that in analysis-based warnings might help with that. https://github.com/llvm/llvm-project/pull/127623