What’s new in glue v0.10?

Below we list some of the main changes in glue v0.10 and in the 3D viewers in glue-vispy-viewers v0.7. As a reminder, you can easily update glue if you are using Anaconda/Miniconda, by doing:

conda install -c conda-forge glueviz

If instead you installed glue with pip, you can update with:

pip install glueviz[all] --upgrade

Note that the 3D viewers (provided by the glue-vispy-viewers plugin package) are now automatically installed when installing glue as above.

In addition to a number of bug fixes and small usability improvements, the following changes are:

Improved linking dialog

The data linking dialog has been redesigned and improved:


In particular, it is now clear in the list of links which components correspond to which datasets. This also fixes previous undesirable behaviors such as components changing names when using the identity link, and such as components being shown alphabetically instead of in their original native order (which has now been fixed). Linking functions can also be grouped by categories.

New data/subset exporters

It is now possible to easily export datasets and subsets by right-clicking (or control-clicking) on them and selecting Export Data or Export Subsets.


Custom data/subset exporters can be easily be defined by users - see Custom Data/Subset Exporters for more details. Currently only a small number of formats are supported by default but this will be expanded in future.

Performance improvements

Performance has been significantly improved (in some cases by factors of 10-100) for cases where 2D datasets were linked with 3D or higher-dimensional datasets, and selections were made in 2D.

Ginga plugin now moved to a separate package

The plugin that allows ginga viewers to be used inside glue has been moved to a new package, glue-ginga. To install this plugin, simply do:

pip install glue-ginga

Compatibility with PyQt5 and Matplotlib 2.x

Glue and the 3D viewers are now fully compatible with PyQt5 and Matplotlib 2.x, which together provide sharper plots on high DPI (e.g. retina) displays.

Creating subset states for categorical components [advanced]

For users who like to create subsets programmatically or in the built-in IPython console, it is now possible to create subset states for categorical components using e.g.:

d.id['source'] == 'name'

Subsets now share more attributes with parent Data objects [advanced]

Subset objects now have properties such as components, visible_components, ndim, shape, and more which are inherited from parent datasets.

Full list of Changes

In addition to the above features, a number of bugs has been fixed since the last release, and a few other small features have been added. A full list of changes can be found in the CHANGES.md file