How to install

This package is available for installation from pypi or conda for the stable version, and directly from source (GitHub’s repository) for the most recent version under development.

pydap can be installed with minimal required dependencies from PyPI via pip as follows:

pip install pydap

Minimal Required Dependencies

The following are minimal dependencies to use pydap as a client.

• python >= 3.10

  • pydap’s CI/CD against stable versions of python (3.10, 3.11, 3.12). pydap recently dropped support for version 3.9.

• numpy >= 2.0

  • pydap recently dropped support for numpy<2.

• scipy

  • it may only be used if netCDF4 (python) library is not currently installed

• requests

  • pydap greatly uses the requests library to connect with OPeNDAP servers through the web, and setup authentication sessions.

Optional dependencies

• matplotlib

• jupyter-lab

• cartopy

• xarray

These are only necessary to run some of the tutorial notebooks.

extra-dependencies

Some extra dependencies can be installed to further exploit pydap’s capabilities. For example, to use pydap as a server, to serve netCDF4 data you can install all the required and extra dependencies as follows:

pip install pydap"[server,netcdf]"

This will install netCDF4-python library as well as all other dependencies to use/run pydap as a lightweight server. With this, pydap implements a lightweight WSGI framework that it can easily be run behind Apache.

To inspect all other possible optional installation with extra dependencies, check the optional-dependencies on the pyproject.toml.

Reproducible environments

We highly recommend using a package installation manager like conda/mamba to install pydap, and any other dependency, in a reproducible and containerized environment. This approach requires having an installation of Miniconda or Anaconda.

The easiest way to install pydap is to use the conda-forge channel. Open a terminal, then run the following commands:

conda create -n pydap -c conda-forge python=3.10 pydap numpy">=2.0" jupyterlab ipython netCDF4 scipy matplotlib

The code above will create a conda environment named “pydap” with many of the commonly used packages for processing and visualizing gridded data, using the latest stable versions (conda release).

To start using pydap, you need to activate the environment with the same name, by running:

conda activate pydap

At this stage, you can install additional packages or remove them if not needed.

Note

If you already have mamba installed, you can replace all conda in the commands with mamba.

PyPI provides additional flexibility for installing python packages that otherwise is hard to achieve with other package installation managers. For example, you can install the latest pydap version directly from the github repository run within the freshly activated pydap environment by running:

pip install --upgrade git+https://github.com/pydap/pydap.git

This version is often not the stable in the sense that it is being actively developed and improved upon by contributors and maintainers of the pydap package.

You can install pydap in developer mode. The developer installation of pydap comes in handy when actively making local changes to pydap (i.e. in your local clone directory), that you are interesting in incorporating into as a new feature, or fixing a bug. This is, the pydap installation gets updated realtime after every change to the code.navigate to a local, cloned pydap repository and run:

pip install -e .

This will install pydap along with the minimal dependencies defined on the pyproject.toml specification. For more on the developer approach of installing pydap, see Contributing to the code.