WIP: Use Ubuntu 16.04 in GitHub actions via docker

[csukuangfj]

Fixes #888 and #854

---

It seems to work.

[csukuangfj]

Will enable it after verification.

[pzelasko]

Woohoo! Thanks!

[csukuangfj]

@pzelasko
Could you try the following commands on your cluster:

conda create -n foo_env
conda activate foo_env
conda install -c k2-fsa -c pytorch k2=1.11 cpuonly pytorch=1.5.0
pip install dataclasses
pip install typing_extensions
python3 -m k2.version

It prints the following for me:

Collecting environment information...

k2 version: 1.11
Build type: Release
Git SHA1:
Git date:
Cuda used to build k2:
cuDNN used to build k2:
Python version used to build k2: 3.6
OS used to build k2: Ubuntu 16.04.7 LTS
CMake version: 3.18.2
GCC version: 5.4.0
CMAKE_CUDA_FLAGS:
CMAKE_CXX_FLAGS:  -D_GLIBCXX_USE_CXX11_ABI=0 -Wno-unused-variable -Wno-strict-overflow
PyTorch version used to build k2: 1.5.0
PyTorch is using Cuda:
NVTX enabled: False
With CUDA: False
Disable debug: True
Sync kernels : False
Disable checks: False

The CUDA versions are still in building and will be available in a few hours.
You can find them at https://anaconda.org/k2-fsa/k2/files once they are built.

[pzelasko]

Thanks, that seems to be working!

$ python3 -m k2.version
Collecting environment information...

k2 version: 1.11
Build type: Release
Git SHA1:
Git date:
Cuda used to build k2:
cuDNN used to build k2:
Python version used to build k2: 3.8
OS used to build k2: Ubuntu 16.04.7 LTS
CMake version: 3.18.2
GCC version: 5.4.0
CMAKE_CUDA_FLAGS:
CMAKE_CXX_FLAGS:  -D_GLIBCXX_USE_CXX11_ABI=0 -Wno-unused-variable -Wno-strict-overflow
PyTorch version used to build k2: 1.5.0
PyTorch is using Cuda:
NVTX enabled: False
With CUDA: False
Disable debug: True
Sync kernels : False
Disable checks: False

[pzelasko]

It also works on a different system with an even older compiler / stdlib (g++ (GCC) 4.8.5 20150623 (Red Hat 4.8.5-28)), that is great. Thanks a lot!

[csukuangfj]

Now there are CUDA versions. Please have a try.