Enable to build the package in normal catkin workspace

[mqcmd196]

In my project, I want to make this package buildable in the catkin workspace. This PR enables it.

I should keep backward compatibility with Docker usage. I'll check whether it is workable with previous usage.

[mqcmd196]

@HiroIshida BTW, what lint command do you execute?

[mqcmd196]

I should keep backward compatibility with Docker usage. I'll check whether it is workable with previous usage.

Done

[mqcmd196]

ERROR: Could not install packages due to an EnvironmentError: [Errno 28] No space left on device: '/tmp/pip-unpacked-wheel-03z28diw/torch/lib/libtorch_cuda_cu.so' -> '/home/user/detic_ws/build/detic_ros/venv/lib/python3.8/site-packages/torch/lib/libtorch_cuda_cu.so'

🤮

[HiroIshida]

@mqcmd196

Thanks. That's really great change

BTW, what lint command do you execute?

I use mypy, isort and flake as listed here:
https://github.com/HiroIshida/detic_ros/blob/master/.github/workflows/peripheral.yml
mypy fails due to broken symbolic link. I'll fix this later, so could you comment out the mypy section in peripheral.yml and let the CI pass for now?

As for no-space-error left error, do you think adding option ARG to dockerfile so that it can skip the jsk_pcl_ros installation, may possibely reduce the volume usage significantly and fix this issue (if it has large dependencies)? Actually I recall that previously set this but it is not used

detic_ros/.github/workflows/docker_build.yml

Line 32 in 7f3419e

INSTALL_JSK_PCL=false

[mqcmd196]

I use mypy, isort and flake as listed here:
https://github.com/HiroIshida/detic_ros/blob/master/.github/workflows/peripheral.yml
mypy fails due to broken symbolic link. I'll fix this later, so could you comment out the mypy section in peripheral.yml and let the CI pass for now?

1784cfb

[mqcmd196]

@HiroIshida

As for no-space-error left error, do you think adding option ARG to dockerfile so that it can skip the jsk_pcl_ros installation, may possibely reduce the volume usage significantly and fix this issue (if it has large dependencies)? Actually I recall that previously set this but it is not used

Before trying it, I added a slightly dirty patch (015e2b3 ). It seems that the test finished correctly.

I think we can make the docker image smaller using the small docker base image. In pip install torch==1.9.0+cu111 torchvision==0.10.0+cu111 torchaudio==0.9.0 part, cuda runtime is installed simultaneously. So, we can use the image here

[HiroIshida]

@mqcmd196 Thanks. I'll consider replacing it.