Module Dependencies

This page describes the different python modules needed by the Jupyter Extension.

TOC

• Modules
• How packages get installed
• Environment Type
• Version Dependencies

Modules

The Jupyter Extension requires these different modules:

Module	What it's used for
jupyter	When there's no ZMQ support, we install the 'jupyter' module
notebook	Jupyter is supposed to come with the notebook module but it can sometimes be missing it. The notebook module is responsible for starting a server and providing the notebook api
ipykernel	The ipykernel module is an implementation of a Jupyter kernel. It's needed to run a python based kernel
nbconvert	nbconvert is a cli tool for converting notebooks to and from different formats. We use this to convert to html and pdf
kernelspec	This isn't really a module. It's a jupyter command as in jupyter kernelspec. If this doesn't work, we reinstall the jupyter module
pandas	Pandas is required to view dataframes (or types that can be converter into a dataframe).
pip	Pip is one of the ways to install packages in Python. It is our most common method for installing
ensurepip	ensurepip is a way to check for pip if it's not found. We check this module to see if we can use it when pip is missing. If this module isn't available, we run a get-pip script.

How modules get installed.

There are number of ways modules can be installed. Almost all of them are similar to pip but depend upon the package manager that the user has.

For local (when the python environment is running on the same machine as the extension host), we do the following based on the environment type:

Environment Type	How an install happens
Default or Unknown	Pip: python -m pip install <module>
Conda	Conda: conda activate <env name> && conda install <module>
VirtualEnv or Venv	<env folder>\scripts\activate && python -m pip install <module>
PipEnv	pipenv install <module>
PyEnv	Not really supported. We just use pip : python -m pip install <module>. Should work though because of how pyenv is setup.
WindowsStore	Same as pip : python -m pip install <module>.
Poetry	<path to poetry exe> add <module>
VirtualEnvWrapper	Same as pip : python -m pip install <module>.
Global	Same as pip : python -m pip install <module>.
System	Same as pip : python -m pip install <module>.

For remote, we cannot detect the environment type so we generally just do this:

%pip install <module>

This is an IPython magic that finds the pip needed to install into a running kernel.

We might be able to detect conda at a later date, and use %conda in that situation.

How is Environment Type determined?

The Python extension does provide the EnvironmentType for every interpreter, but we also make sure that we can run the associated package manager before we run an install. If we can't, everything defaults back to pip.

Version Dependencies

Each of the modules we install have a set of versions that we support because they modify behavior of the module.

Jupyter

Jupyter isn't really a module, but a series of submodules. Their are two important to us - notebook and nbconvert

notebook

Version	Why	Do we verify
5.0	Oldest we tested.	We don't check version here, but we probably should.
6.4	Main release in use.	No

IPython

Version	Why	Do we verify
7.0	Oldest we tested	We don't check version here, but we probably should.
8.0	Required for notebook debugging support.	We don't currently upgrade if below this, but instead put up a warning message
8.4	With Python 3.10, this is required for full debugging support because of bugs in the implementation. With Python 3.9, version 8.0 will work.	Not checked

pandas

Version	Why	Do we verify
0.20	Required for DataFrame viewing because of ability to generate JSON	We ask the user to install a newer version if not found. Works while debugging too.

nbconvert

Version	Why	Do we verify
6.0.0	Oldest tested.	Not verified

traitlets

Version	Why	Do we verify
5.1.1	There was a known issue with this package.	Verified when a specific error message comes up. See error checking