The manual installation instructions use conda to manage the required dependencies. Alternatively we could build RPMs for installation using yum, which may be more familiar to admins. This is complicated by a few things:
Different JupyterHub configurations may want different dependencies. One way around this would be to bundle potential expected dependencies all together (e.g. include several different potential authenticators) into a single RPM.
The user's notebook environments will still need to be created and hosted somewhere. Should we provide an RPM for a standard notebook environment? Or leave it up to the user to create these externally?
The manual installation instructions use conda to manage the required dependencies. Alternatively we could build RPMs for installation using
yum
, which may be more familiar to admins. This is complicated by a few things:Different JupyterHub configurations may want different dependencies. One way around this would be to bundle potential expected dependencies all together (e.g. include several different potential authenticators) into a single RPM.
The user's notebook environments will still need to be created and hosted somewhere. Should we provide an RPM for a standard notebook environment? Or leave it up to the user to create these externally?