mizvekov wrote: > Wonder what test should I add here. This only affects the content of the > internal folding set so we always create new types but result is still > correct.
One possibility is doing an AST test, there are some examples in there that show how you can test two different AST nodes have the same address. The best here would be some kind of performance regression test. We don't have anything like that per-se in the test suite unfortunately. But if you can isolate a good test case, you can make it a normal regression test, which with the fix is still within CI limits of time and space, but which pushes much farther out without the fix. This would not be without precedent, there already exists tests like that. https://github.com/llvm/llvm-project/pull/79182 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
