New posts in error-checking

What is the canonical way to check for errors using the CUDA runtime API?

cuda error-checking