Why is JupyterLab version 0.35.4 the standard version e.g. in Anaconda?


I’m trying to ease the process of notebook deployment in a classroom environment. To me, JupyterLab seems to offer a more friendly UX than the ascetic UI of a standard Jupyter Notebook. However, widgets don’t work for me, neither in my local Anaconda environment (latest, using Jupyterlab 0.35.4) nor in Microsoft Azure Notebooks and this is a well known problem. When run as a standard Jupyter notebook, everything’s fine.

My question is: Why is such a dated (2015?) version the standard? There seems to be a release frenzy (> 4000 releases as of today!) but this doesn’t reach downstream distributions. What is the reason behind this?



Jupyterlab 0.35.4 is the latest released version of Jupyter Lab as far as I can tell from PyPI: https://pypi.org/project/jupyterlab/#history

It was released in November 2018, are you sure you installed this version? The 2015 seems off.

Often the packaging in the many, many downstream package managers is done by someone not part of the core team. I don’t know what the arrangement in the Jupyter Lab team is. I typically use the following assumptions for a Python package:

  • PyPI is updated by the core team when there is a new release, this is the single source of truth.
  • conda-forge packages are maintained by the community, sometimes also by the core team. If actively maintained these probably lag PyPI by a few days
  • anaconda packages are maintained by Anaconda Inc, I have no idea how, if, when these get updated
  • debian, ubuntu, etc packages often don’t exist, are not used a lot by the Python community, might be maintained by someone the lag wrt PyPI is highly variable

A good thing about conda-forge packages is that anyone (including users) can trigger an update once there is a new PyPI package. For many packages this is a “trivial” step that is highly automatised, but (I think) still requires a human.



You’re right, I must have misread something in the conda messages - thanks!