One release here, one release there, pretty soon you have ten!

This release incorporates several new backend features that make building faster and more reproducible, in addition to the usual updates of versions and new libraries. In particular:

• The Python gds environment is now built from scratch, rather than added on top of the base environment provided by minimal-notebook. This makes resolving the versions a lot faster and does not create conflicts with some libraries as in the past.
• The gds environment is automatically turned on in the container, so the user should see no difference with the past model in accessing geo libraries
• The gds environment is built from a single .yml file that includes all downloads (also from pip), and which can thus be used to recreate the environment in other contexts if necessary
• The python and R kernels are renamed to include the version of the GDS env and also point to the appropriate environment. The base kernel that ships with minimal-notebook is hidden to avoid confusion (though the environment itself is present in the container).

Main additions as detailed in #80

Citing

@software{gds_env,
  author = { { Dani Arribas-Bel } },
  title = {\texttt{gds\_env}: A containerised platform for Geographic Data Science},
  url = {https://github.com/darribas/gds_env},
  version = {10.0},
  date = {2023-10-24},
}

A book and an arm

Minor point release that only includes as addition files to build explicit conda environments in the three previously supported platforms (linux/macOS intel/windows) and macOS arm (aka Apple silicon) that can run the 1.0 version of the GDS Book. No updates to the platform.

Bookworm

Update of the stack:

• Main changes available in #76
• Bash kernel added in gds_dev
• Version to support the official release of the GDS Book
• Explicit files have been copied from Github Actions after successful completion (linux and macOS), and reproduced locally for Windows (without pygeoda as it requires a large C++ compiler install) and uploaded manually in 28892d8
• Likely the last release with pandas 1.X

Tag early, release later

This release provides an update of versions of all core packages, and the following advances:

• The main infrastructure addition in this release is a set of explicit files (gds_py_explicit_XXX-latest.txt, available here) to recreate the exact Python environment in Linux, MacOS, and Windows (all intel-only, for now). These are created following the cloning guidance in conda, and can be replicated running conda create/install --name myenv --file gds_py_explicit_XXX-latest.txt
• CI has also been expanded to include a re-build (upon success) of the explicit gds_py_explicit_XXX-latest.txt files on each commit
• Main additions/removals as specified in #73

v7.0 - Spooky exploration

Autumn release updating the stack to most recent versions. Most notably:

• geopandas 0.10.2 with interactive mapping through gdf.explore()
• pysal 2.5
• XYZservices to unify basemap providers
• contextily 1.2 with XYZservices backend
• Parquet support in R for spatial data through sfarrow

Full list of version differences is available here (Python) and here (R)

Citing

@software{gds_env,
  author = { { Dani Arribas-Bel } },
  title = {\texttt{gds\_env}: A containerised platform for Geographic Data Science},
  url = {https://github.com/darribas/gds_env},
  version = {7.0},
  date = {2019-08-06},
}

v6.1 - Easter Egg

Point release fixing a few regressions introduced in 6.0 and other working issues that cropped up on first use. Upgrade from 6.0 is recommended. Issues and progress was tracked on Milestone 6.1

Regressions fixed

• jupyterbook is now again part of the base environment so it can be used in tandem with the rest of the python stack
• decktape is installed from sources and now works as expected
• texbuild install is updated to point to specific Python version so it works again

Other additions

• Experimental version of geopandas_view added
• Alpha release of dask-geopandas included
• Pinning to latest version of tobler (ahead of PySAL version)

Citing

@software{gds_env,
  author = { { Dani Arribas-Bel } },
  title = {\texttt{gds\_env}: A containerised platform for Geographic Data Science},
  url = {https://github.com/darribas/gds_env},
  version = {6.1},
  date = {2019-08-06},
}

v6.0 - Divide and conquer

This release updates each stack significantly (see detailed changes), and provides several additional infrastructure innovations:

• Upgrade to JupyterLab 3.0 (through minimal-notebook
• Drop of qgrid and KeplerGL, at least temporarily while the projects become compatible with JupyterLab 3.0
• Conda installs relating to web development (Jupyter-book, Jekyll, pyppeteer, etc.) have been removed from gds_py and are now included in a separate conda environment (dev) on gds_dev. To access them, conda activate dev inside gds_dev.
• Switch from MKL Blas to OpenBLAS on the gds_py stack
• Taken the changes above, gds_py is not just over 3.5GB in footprint, down from over 6GB in 5.0
• Versions of packages in gds_py are hardcoded so the stack stays stable over time
• CI testing of gds_py pins to versionned environment files

Citing

@software{gds_env,
  author = { { Dani Arribas-Bel } },
  title = {\texttt{gds\_env}: A containerised platform for Geographic Data Science},
  url = {https://github.com/darribas/gds_env},
  version = {6.0},
  date = {2019-08-06},
}

Corona-ready

This release updates each stack significantly (see detailed changes), and provides several additional infrastructure additions to the project:

• New website at https://darribas.org/gds_env
• Additional build and install guides for Docker and Virtualbox
• Binder badge
• Each stack is now in its own folder within the repository
• CI testing of gds_py now includes also libraries installed through pip
• New diagram:

Citing

@software{gds_env,
  author = { { Dani Arribas-Bel } },
  title = {\texttt{gds\_env}: A containerised platform for Geographic Data Science},
  url = {https://github.com/darribas/gds_env},
  version = {5.0},
  date = {2019-08-06},
}

A little one for the tiles

Point release to include the 1.0 release of contextily. In addition, the following updates are included too:

• Some updates to individual packages (see diff here)
• GeoJSON plugin for JupyterLab
• Adding cenpy to gear for 5.0

Citing

@software{gds_env,
  author = {{Dani Arribas-Bel}},
  title = {\texttt{gds\_env}: A containerised platform for Geographic Data Science},
  url = {https://github.com/darribas/gds_env},
  version = {4.1},
  date = {2019-08-06},
}

More Geo, less bulk

This version adds a new flavour of the gds_env containers, gds_dev, which offloads all dev tools from the other stacks, and adds a few other ones. There are also some changes in the list of libraries included (less non-geo, a few more geo). Important additions/removal followed #25, and a few other issues were also closed (#18, #24).

• Specific list of Python libraries is available on stack_py.txt and the detailed changelog is available as a diff.
• Specific list of R libraries is available on stack_r.txt and the detailed changelog is available as a diff.

Installation

• Python stack only:

docker pull darribas/gds_py:4.0

• Full stack: Python + R:

docker pull darribas/gds:4.0

• Full stack + development tools:

docker pull darribas/gds_dev:4.0

Citing

@software{gds_env,
  author = {{Dani Arribas-Bel}},
  title = {\texttt{gds\_env}: A containerised platform for Geographic Data Science},
  url = {https://github.com/darribas/gds_env},
  version = {4.0},
  date = {2020-02-26},
}