Well, this is only "a great swap" in cases where the time taken is already so high that you won't notice a 10x.

But this tradeoff would actually pay off where the compile time has a similar improvement as the size.