FindCUDA: run_nvcc.cmake doesn't have a place to slip ccache in
I'm trying to use ccache with nvcc (CUDA) to speed up compilation of OpenCV.
In #16953 (closed) and !949 (merged), CUDA_COMPILER_LAUNCHER
was added but as far as I understand, this doesn't affect to run_nvcc.cmake.
As a result, if a project uses run_nvcc.cmake (like OpenCV), nvcc will not be invoked via ccache.
First of all, is my understanding correct ? Second, is it possible to ask upgrade of run_nvcc.cmake, so it'll try to use ccache with nvcc ?
Edited by Brad King