ChuanqiXu9 wrote: > I see, so because we no longer allow inconsistent options when importing, we > now issue a lot of error diagnostics. But for many of the options, > differences between TUs should have no impact on modules anyway, and so those > errors are false positives. Is that about correct?
Yes. > If I'm right, then what about a different way forward: update the individual > options to specify whether it is an error, a warning, or silently fine for > the option to be inconsistent across module boundaries, the emit diagnostics > (or not) as appropriate for each option. It's more up-front work, but it > means we stop issue false positive errors without losing diagnostic fidelity. > (Note, we may find we want to reorganize Options.td so it's easier for us to > specify "this entire block of options are ones for which module inconsistency > is an error".) The problem may be about the amount of work. Although I am not sure how far we can make it in practice, it is technically doable. --- Some other thoughts about false-positive error diagnostics. I am wondering, if we don't want false-positive error diagnostics at all, then we have to do what I proposed in this PR (to downgrade errors to warnings for inconsistent flags). Otherwise, we can always construct at least one example that it is false-positive to diagnose the inconsistency. e.g., ``` /// m1.cppm export module m1; // eof // m2.cpp import m1; ``` then is it really matter to compile `m1.cppm` with `-std=c++23` and import it to `m2.cpp` which is compiled as `-std=c++20`? I don't think so as far as I know. So is it a false-positive error diagnostic? I think yes. And `-std=` may be the most impactful options, if we can construct cases for `-std=`, then we can construct cases for other flags too. Then, if we can't tolerant false positive error diagnostics, we should emit warnings instead for all the flags. https://github.com/llvm/llvm-project/pull/117840 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
