jupyterlite_xeus

JupyterLite loader for Xeus kernels

Requirements

• JupyterLab >= 4.0.0

Install

To install the extension, execute:

pip install jupyterlite_xeus

Usage

From environment.yaml

xeus-python kernel

To create a xeus-python kernel with a custom environment, one creates an environment.yaml file with xeus-python and the desiered dependencies. Here is an example with numpy as a additional dependency:

name: xeus-lite-wasm
channels:
  - https://repo.mamba.pm/emscripten-forge
  - conda-forge
dependencies:
  - xeus-python
  - numpy

To build the jupyterlite run the following command, where environment.yaml is the path to the file you just created

jupyter lite build --XeusAddon.environment_file=some_path/to/environment.yaml

xeus-lua / xeus-sqlite / xeus-<mylang>

To create a xeus-lua or xeus-sqlite kernel you can do the same as above, just with

dependencies:
  - xeus-lua

or

dependencies:
  - xeus-sqlite

Note that xeus-sqlite and xeus-lua do not yet support additional dependencies. To build the jupyterlite, run again:

jupyter lite build --XeusAddon.environment_file=environment.yaml

Multiple kernels

To create a deployment with multiple kernels, you can just add them to the environment.yaml file:

name: xeus-lite-wasm
channels:
  - https://repo.mamba.pm/emscripten-forge
  - conda-forge
dependencies:
  - xeus-python
  - xeus-lua
  - xeus-sqlite
  - numpy

From local environment / prefix

When developing a xeus-kernel, it is very usefull to be able to test it in a jupyterlite without having to publish it to emscripten-forge. Therefore you can also use a local environment / prefix to build a jupyterlite with a custom kernel.

Create a local environment / prefix

This workflow usually starts with creating a local conda environment / prefix for emscripten-wasm32 with all the dependencies you need to build your kernel (here we install dependencies for xeus-python )

micromamba create -n xeus-python-dev \
    --platform=emscripten-wasm32 \
    -c https://repo.mamba.pm/emscripten-forge \
    -c https://repo.mamba.pm/conda-forge \
    --yes \
    "python>=3.11" pybind11 nlohmann_json pybind11_json numpy pytest \
    bzip2 sqlite zlib libffi xtl pyjs \
    xeus xeus-sqlite

Build the kernel

This dependes on your kernel but will look smth like this:

# path to your emscripten emsdk
source $EMSDK_DIR/emsdk_env.sh

WASM_ENV_NAME=xeus-python-dev
WASM_ENV_PREFIX=$MAMBA_ROOT_PREFIX/envs/$WASM_ENV_NAME

# let cmake know where the env is
export PREFIX=$WASM_ENV_PREFIX
export CMAKE_PREFIX_PATH=$PREFIX
export CMAKE_SYSTEM_PREFIX_PATH=$PREFIX

cd /path/to/your/kernel/src
mkdir build_wasm
cd build_wasm
emcmake cmake \
    -DCMAKE_BUILD_TYPE=Release \
    -DCMAKE_FIND_ROOT_PATH_MODE_PACKAGE=ON \
    -DCMAKE_INSTALL_PREFIX=$PREFIX
    ..
emmake make -j8 install

When running jupyter lite build we pass the prefix options and point it to the local environment / prefix we just created:

jupyter lite build  --XeusAddon.prefix=$WASM_ENV_PREFIX

Mounting additional files

To copy additional files and directories into the virtual filesytem of the xeus-lite kernels you can use the --XeusAddon.mount option. Each mount is specified as a pair of paths separated by a colon :. The first path is the path to the file or directory on the host machine, the second path is the path to the file or directory in the virtual filesystem of the kernel.

jupyter lite build \
    --XeusAddon.environment_file=environment.yaml \
    --XeusAddon.mounts=/some/path/on/host_machine:/some/path/in/virtual/filesystem

Contributing

Development install from a conda / mamba environment

Create the conda environment with conda/mamba/micromamba (replace micromamba with conda or mamba according to your preference):

micromamba create -f environment-dev.yml -n xeus-lite-dev

Activate the environment:

micromamba activate xeus-lite-dev

python -m pip install -e .   -v --no-build-isolation

Packaging the extension

See RELEASE