--reject is nice, but I often like to try --3way first.

(Use case is using stg to load patchsets from QEMU mailing list as a stack for review purposes. Sometimes the patches are slightly out of date and --3way or --reject is needed to get the patches loaded.)

I could likely write a patch to just "plumb" a --3way option through, but I was wondering if a more general purpose solution for passing "arbitrary" arguments to git apply would be wanted instead, or if "the stupid approach" here was actually best. It might also be nice, because mbox files can have multiple patches within them, to allow the --interactive mode, too.

Please advise!
--js