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
I would like to propose that work be done to eliminate the need for jhub-native-proxy to be installed in the application environment.
Consider the standard JupyterHub session for a moment. Thanks to nb_conda_kernels, the kernel environments where notebooks are run do not need to be the same as the JupyterLab environment responsible for rendering the JupyterLab UI. This separation is beneficial not just because it allows us to leverage multiple environments in a single JupyterLab session. But it also ensures that JupyterLab's dependencies don't conflict with the dependencies in the kernel environment. The packages like ipykernel and r-irkernel are far more lightweight in that regard.
For application, jhub-native-proxy sits in the position of JupyterLab, and the application code sits in the position of the kernel environment. jhub-apps should at least optionally allow these two environments to be separate.
Value and/or benefit
For me, the most important benefit of this improvement is that it will make it easier to port applications from other data science environments.
Anything else?
No response
The text was updated successfully, but these errors were encountered:
Feature description
I would like to propose that work be done to eliminate the need for
jhub-native-proxy
to be installed in the application environment.Consider the standard JupyterHub session for a moment. Thanks to
nb_conda_kernels
, the kernel environments where notebooks are run do not need to be the same as the JupyterLab environment responsible for rendering the JupyterLab UI. This separation is beneficial not just because it allows us to leverage multiple environments in a single JupyterLab session. But it also ensures that JupyterLab's dependencies don't conflict with the dependencies in the kernel environment. The packages likeipykernel
andr-irkernel
are far more lightweight in that regard.For application,
jhub-native-proxy
sits in the position of JupyterLab, and the application code sits in the position of the kernel environment.jhub-apps
should at least optionally allow these two environments to be separate.Value and/or benefit
For me, the most important benefit of this improvement is that it will make it easier to port applications from other data science environments.
Anything else?
No response
The text was updated successfully, but these errors were encountered: