steakhal wrote: > > I'm a bit unsatisfied with the way of testing this. > > IMO the `getIssueString` should be tested in a unittest, because it's nice > > and pure. > > How about creating a new unittest as > > `clang/unittests/Analysis/IssueHashTest.cpp` where we would assert that for > > a given Decl in a C++ snippet it produces the expected issue hash string. > > I personally don't see the value in this more complicated testing. The actual > value of a hash is irrelevant (and should not be pinned), the only important > property is that we assign different hashes to different bugs -- and that is > clearly tested in the existing tests.
This is not true. `getIssueString` is the string that gets hashed. It's human-readable. It is way more direct than counting the number of files. And I believe that having the setup code (LIT RUN lines) duplicated for 4 tests is more than setting up a unittest. https://github.com/llvm/llvm-project/pull/214102 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
