CUDA compilation is broken

[gpanders]

After commit 54766b9 OpenMVS fails to build with the following error:

CMake Error: The following variables are used in this project, but they are set to NOTFOUND.
Please set them or make sure they are set and tested correctly in the CMake files:
CUDA_CUDA_LIBRARY (ADVANCED)
    linked by target "MVS" in directory /src/openmvs/libs/MVS

Reverting 54766b9 fixes the issue and allows OpenMVS to compile normally.

I am building OpenMVS in a Docker container based on the NVIDIA CUDA Docker image.

[cdcseacave]

since CUDA_CUDA_LIBRARY is not set only if if(CUDA_FOUND) is true, then there is a problem with the CMake module which should set CUDA_CUDA_LIBRARY. Pls make sure you have the latest version of CMake. Ubuntu 16.04 is compiling fine with the default CMake from apt-get.

[gpanders]

I am using Ubuntu 16.04 with CMake version 3.5.1 from apt-get

[gpanders]

To anyone else who may stumble upon this problem, the solution was to add the following CMake flag -DCMAKE_LIBRARY_PATH=/usr/local/cuda/lib64/stubs/

See here for more info.

[cdcseacave]

thx @gpanders