I was able to accomplish this using the exact method that spirilis described using a hot plate. I did however have a couple short out during testing because it's very difficult cleaning out any leftover flux from underneath them. That's another pitfall to watch out for.
I'd also agree that using long traces would be a good fallback plan.