sepavloff added a comment. In D134337#3817878 <https://reviews.llvm.org/D134337#3817878>, @mgorny wrote:
> Ok, how about this variant? I think it's the simplest I can come up with that > roughly matches the old behavior and adds what's necessary for the new. > > The algorithm is to use: > > 1. `<triple>-<mode>.cfg` if available, using the real `<mode>`, > 2. `<triple>-<mode>.cfg` if available, using `<mode>` inferred from > executable name, > 3. `<triple>.cfg` + `<mode>.cfg` if either is available, using the real > `<mode>`, > 4. `<triple>.cfg` + `<mode>.cfg` if either is available, using the `<mode>` > inferred from executable. This algorithm looks good. It is compatible with the existing algorithm if target is not overridden. In the case of overridden target behavior is different, but you a right, using `x86_64.cfg` in compilation for 32-bit target is not natural. Probably we should document the difference in `ReleaseNotes.rst`. Also documentation in `UserManual.rst` should be updated. CHANGES SINCE LAST ACTION https://reviews.llvm.org/D134337/new/ https://reviews.llvm.org/D134337 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits