[PATCH] D13144: [CUDA] propagate to CUDA sub-compilations target triple of opposite side.

[Eric Christopher via cfe-commits]

echristo added a comment.

Would this make more sense to put in the Compilation class rather than on the Action? It seems like it could go next to DefaultToolchain or could even be a separate toolchain itself so that all of the bits could be looked up?

Thanks!

-eric


http://reviews.llvm.org/D13144