Hi Richard,
Just as a suggestion, maybe we'd want to extend this
to other intrinsics in future so a -fno-inline-intrinsic=pack[,...]
is more future proof? (I'd inline all intrinsics by default thus
only provide the negative form). You can avoid the extra
option parsing complexity by only literally adding
-fno-inline-intrinsic=pack for now.
I agree that such an option would make sense, I think this is
something we should consider for gcc 11.
In this instance, your reply shows that the option is poorly named,
because it is actually not about the PACK intrinsic, but the internal
packing that happens for arguments.
Maybe -finline-repack would be a better name? -finline-internal-pack?
Regards
Thomas
