We're getting a fairly large collection of Python dependencies, and we've occasionally hesitated to add some useful module because of the extra dependency that it would introduce. This is what docker containers are for: you can put everything into a container, then somebody just runs it with docker run <container>. Configuring docker isn't trivial the first time but then you never have to deal with it again, and it can pull the container down fresh every time - so we can update it with point releases or bug fixes.
We're getting a fairly large collection of Python dependencies, and we've occasionally hesitated to add some useful module because of the extra dependency that it would introduce. This is what docker containers are for: you can put everything into a container, then somebody just runs it with
docker run <container>
. Configuring docker isn't trivial the first time but then you never have to deal with it again, and it can pull the container down fresh every time - so we can update it with point releases or bug fixes.