You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
The current setup process for RAI is challenging, particularly when it comes to resolving Python compatibility issues and dependencies. Users who attempt to set up RAI in alternative environments, such as Conda or standalone Python configurations, face significant hurdles, leading to excessive time spent troubleshooting. This creates a barrier to integrating RAI with other frameworks like LeRobot.
Describe the solution you'd like
A Docker image or a simplified Dockerfile should be provided to enable seamless RAI setup. This image should mirror the tested environment used in RAI's CI pipelines, ensuring compatibility and reliability. Hosting the built Docker image on a public registry (e.g., Docker Hub) would further simplify the process for users by allowing them to pull and use the image directly.
Since RAI's dependencies change often, the hosted docker image should be updated often, perhaps using Jenkins or other solutions
Describe alternatives you've considered
Detailed step-by-step guides for alternative environments (e.g., Conda or standalone Python setups).
Additional context
This Docker-based solution would significantly reduce the time and effort required for users to get started with RAI. A public Dockerfile would also aid advanced users who wish to adapt the environment for their unique setups. Including this Docker image in RAI's official documentation would enhance user accessibility and streamline the onboarding process.
The text was updated successfully, but these errors were encountered:
@vmayoral
Effort started: #377
The images are working, however there are still some decisions to be made e.g.:
Mounting current folder as /rai overwrites the /rai folder used to install deps and build packages in Dockerfile. This is not optimal, as with first docker run there are no build files which require an additional command to be ran (colcon build...)
Publish target for images: ghcr vs docker registry
We have merged the changes. While the solution is far from perfect and hasn’t been tested as thoroughly as Poetry RAI, we hope it will suffice for the tryouts. #377
Is your feature request related to a problem? Please describe.
The current setup process for RAI is challenging, particularly when it comes to resolving Python compatibility issues and dependencies. Users who attempt to set up RAI in alternative environments, such as Conda or standalone Python configurations, face significant hurdles, leading to excessive time spent troubleshooting. This creates a barrier to integrating RAI with other frameworks like LeRobot.
Describe the solution you'd like
A Docker image or a simplified Dockerfile should be provided to enable seamless RAI setup. This image should mirror the tested environment used in RAI's CI pipelines, ensuring compatibility and reliability. Hosting the built Docker image on a public registry (e.g., Docker Hub) would further simplify the process for users by allowing them to pull and use the image directly.
Since RAI's dependencies change often, the hosted docker image should be updated often, perhaps using Jenkins or other solutions
Describe alternatives you've considered
Additional context
This Docker-based solution would significantly reduce the time and effort required for users to get started with RAI. A public Dockerfile would also aid advanced users who wish to adapt the environment for their unique setups. Including this Docker image in RAI's official documentation would enhance user accessibility and streamline the onboarding process.
The text was updated successfully, but these errors were encountered: