Martin =?utf-8?q?Storsjö?= <[email protected]> Message-ID: In-Reply-To: <llvm.org/llvm/llvm-project/pull/[email protected]>
erichkeane wrote: > Thanks for looking at this! > > > Otherwise, what is the behavior if someone specifies BOTH `ms_struct` AND > > `gcc_struct`? It seems to me we should both test and define what that means. > > On one hand, I'm not entirely sure if we want to explicitly document the > behaviour for this case - should we rather say that specifying both is > unspecified? Then we can make that into an error (due to the user passing > directly conflicting attributes) later if we want to? > > > At the moment, it seems like this would just default to the `gcc_struct`, > > which seems wrong to me. > > What would be the right behaviour to you? If both are passed, I don't really > see which outcome would be less wrong, other than erroring out? > > I'll try to expand the documentation briefly about what the differences are. IMO, just error'ing is the only right behavior I could think of. I left it open in case you had a better idea. Leaving it 'undefined'/'unspecified' is pretty unacceptable. Defaulting to `gcc_struct` seems arbitrary and without any reason whatsoever, and I don't see any value to that. Barring any good ideas, this needs to error. https://github.com/llvm/llvm-project/pull/71148 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
