If you're frustrated with Docker's slow pace of development are are looking for a good alternative, have a look at Singularity containers. They're interoperable with Docker Hub and OCI compatible images, and offer a much better experience for HPC, machine learning and big data environments. Their GPU support is top notch. They don't run on as many platforms as Docker, but are aiming to grow into cross-platform. Sylabs is a small team in Austin, TX that does a lot of the development work, but they are quickly gaining adoption in academia. It's available on Cedar and Graham in Compute Canada (a large supercomputing cluster that serves most of Canada). We use Singularity on all our lab machines as it is much easier for IT admins to manage, and doesn't require sudo access. Their GPU support is top notch.