Lambo-IITian wrote: > > I reviewed the failing tests and they all come from Clang AST JSON dump > > checks where the expected `"offset"` values do not match the actual output. > > AST structure itself is correct , only the recorded byte offsets differ > > slightly. > > These tests validate exact source location metadata, which is sensitive to > > environment differences (e.g., line ending normalization). So the failures > > are location mismatches, not compiler logic issues. > > What kind of "offset" issues are you encountering with after this patch? And > why are there offset issues if the change should only aim to fix the > qualifier problem?
This change makes the comparison check both the template name and qualifiers, so differently qualified types are no longer treated as the same. That doesn’t change the semantic AST, but it can shift which internal node gets associated with a source location. Since the JSON tests check exact byte offsets, this shows up as small `"offset"` differences. So the failures are just due to updated location metadata. I’ll update the test expectations and rerun the suite to make sure everything passes. https://github.com/llvm/llvm-project/pull/180175 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
