steakhal wrote: > > I've not measured any of this. It may break stuff because > > checkers/heuristics may only expect Conjured symbols at places where we > > would pass invalidation artifact symbols instead - breaking those. You can > > see a bunch of places adopted e.g. iterator checkers. > > My immediate reaction is that if this "mirrors" Conjured symbols, then why > cannot this be a subtype of `SymbolConjured`? Did you think about this option > and reject it for some reason? > > (I didn't read the PR yet, there may be a good reason.)
I think they are orthogonal, and is-a relationship wouldn't be correct there. It would certainly make the migration easier, but IMO it wouldn't be a great fit. https://github.com/llvm/llvm-project/pull/207155 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
