Installation

Requirements

  • Python 3.9 or newer.

  • git-annex, as for DataLad itself. Install it with your system’s package manager (e.g. sudo apt-get install git-annex on Debian/Ubuntu, brew install git-annex on macOS), from conda-forge (conda install -c conda-forge git-annex), or into your Python environment with pip install git-annex. The DataLad Handbook covers more options.

  • FUSE, only for mounting datasets with datalad fusefs. The mfusepy bindings used by datalad-fuse work with the FUSE 3 or the FUSE 2 library, together with its fusermount3 or fusermount helper. On Debian and Ubuntu:

    sudo apt-get install fuse3

    (sudo apt-get install fuse on releases without the fuse3 package). If both libraries are installed, FUSE 2 is used; set the environment variable FUSE_LIBRARY_NAME=fuse3 to use FUSE 3 instead.

    FUSE mounts are developed and tested on Linux. Opening files from Python (see Python usage) and datalad fsspec-head do not need FUSE at all.

Installing datalad-fuse

Install the latest release from PyPI with pip; DataLad, fsspec and the other Python dependencies are installed along with it:

$ python3 -m pip install datalad-fuse

To install the development version, which may include features that are not released yet:

$ python3 -m pip install git+https://github.com/datalad/datalad-fuse.git

Checking the installation

datalad-fuse is a DataLad extension: it adds commands to DataLad. Check that DataLad finds them and that git-annex is available:

$ datalad fusefs --help
Usage: datalad fusefs [-h] [-d DATASET] [-f] [--mode-transparent]
...
$ git annex version
git-annex version: 10.20260901
...

and that the Python package can be imported:

$ python3 -c "import datalad_fuse; print(datalad_fuse.__version__)"