Poetry tips and tricks

Mount an ipykernel in a poetry environment to use with your global Jupyter

Install ipykernel as development dependency:

poetry add -D ipykernel

Register the ipykernel:

poetry run python -m ipykernel install --user --name my_kernel

Check that the kernel is registered:

jupyter kernelspec list

Install the main package in the ipykernel so that it’s available from anywhere

Poetry by default will try to do it, but Jupyter will not recognize this. You need to add a setup.py alongside your __init__.py:

from setuptools import setup, find_packages

setup(
    name='name_of_your_package',
    version='0.1.0',
    packages=find_packages(include=['name_of_your_package', 'name_of_your_package.*'])
)

See the location of the current venv:

poetry env info --path

Managing different poetry environments in different branches

By default, Poetry will create one global environment for a given project name. This can be problematic when you use different dependencies on different branches, but within the same project.

Solution 1: recreate the environment at switch

Switch branches and recreate the environment from scratch:

poetry env list
poetry env remove env_name_like_listed_above
poetry lock
poetry install 

or

poetry install --remove-untracked

Solution 2: keep branches in separate directories and make poetry create local environments

Clone each branch into a separate local directory. To make poetry create separate environments w/o having to rename the project name in pyproject.toml, run:

• for all the projects:

poetry config virtualenvs.in-project true

• only for those projects - run separately in each directory

poetry config virtualenvs.in-project true --local

And then run:

poetry lock 
poetry install

Lock is important to create a new venv.

To see the current config:

poetry config --list

src1 src2 src3

Then mount Ipython kernel in each environment separately.