Skip to content

Latest commit

 

History

History
123 lines (81 loc) · 3.03 KB

install.md

File metadata and controls

123 lines (81 loc) · 3.03 KB
Raw

gch4i guide installation

Prerequisites

Create environment

conda env create -f environment.yml
activate gch4i

or

mamba env create -f environment.yml
activate gch4i

The packages necessary to run the project are now installed inside the conda environment.

Note: The following sections assume you are located in your conda environment.

Set up project's module

To move beyond notebook prototyping, all reusable code should go into the gch4i/ folder package. To use that package inside your project, install the project's module in editable mode, so you can edit files in the gch4i folder and use the modules inside your notebooks :

pip install --editable .

To use the module inside your notebooks, add %autoreload at the top of your notebook :

%load_ext autoreload
%autoreload 2

Example of module usage :

from gch4i.utils.paths import data_dir
data_dir()

Set up Git diff for notebooks and lab

We use nbdime for diffing and merging Jupyter notebooks.

To configure it to this git project :

nbdime config-git --enable

To enable notebook extension :

nbdime extensions --enable --sys-prefix

Or, if you prefer full control, you can run the individual steps:

jupyter serverextension enable --py nbdime --sys-prefix

jupyter nbextension install --py nbdime --sys-prefix
jupyter nbextension enable --py nbdime --sys-prefix

jupyter labextension install nbdime-jupyterlab

You may need to rebuild the extension : jupyter lab build

Set up Plotly for Jupyterlab

Plotly works in notebook but further steps are needed for it to work in Jupyterlab :

  • @jupyter-widgets/jupyterlab-manager # Jupyter widgets support
  • plotlywidget # FigureWidget support
  • @jupyterlab/plotly-extension # offline iplot support

There are conflict versions between those extensions so check the latest Plotly README to ensure you fetch the correct ones.

jupyter labextension install @jupyter-widgets/jupyterlab-manager@0.36 --no-build
jupyter labextension install plotlywidget@0.2.1  --no-build
jupyter labextension install @jupyterlab/plotly-extension@0.16  --no-build
jupyter lab build

Invoke command

We use Invoke to manage an unique entry point into all of the project tasks.

List of all tasks for project :

$ invoke -l

Available tasks:

  lab     Launch Jupyter lab

Help on a particular task :

$ invoke --help lab
Usage: inv[oke] [--core-opts] notebook [--options] [other tasks here ...]

Docstring:
  Launch Jupyter lab

Options:
  -i STRING, --ip=STRING   IP to listen on, defaults to *
  -p, --port               Port to listen on, defaults to 8888

You will find the definition of each task inside the tasks.py file, so you can add your own.