1.9.0

• Added a CHANGELOG.txt to the repo noting changes that significantly change usage concerns
• Implemented a new (optional) way of managing storage that doesn't use the working directory (which is not well-defined for notebooks). You can enable this with an environment variable. Larger changes to storage are in the works.
• Clean up code around type() and isinstance() to better handle future changes to Python
• Smaller changes around string interpolation in some parts of the code
• Remove some more old python2 compatibility measures
• Progress bars (enable with an env var) thanks to EricThomson
• h5py pin for tensorflow

Please report any issues relating to storage paths, and please try the new env var.

1.8.9

This release moves us to python 3.7 and 3.8 being the supported python revisions for CaImAn, and also (because of upstream packaging changes in conda) marks a move to Tensorflow 2.2+ as our supported version. All diffs after this release may introduce dependencies on Python 3.7 and Tensorflow 2.2+.

Other changes:

• Made intro doc more focused on software installation
• Added reference to another volpy paper
• Updated setup.py to add current contact information
• Removed software pins that were no longer necessary
• update_temporal_components() - small improvements in docs

1.8.8

• Improved error handling for input
• Improvements to demo_dendritic
• VolPy paper datafiles
• Add exceptions to catch problems in background extraction
• small adjustments to perf guide
• other volpy improvements
• Remove obsolete demos
• Fix serialisation issues with motion correction init code

1.8.7

• Updates to volpy gui, add simple_thresh function
• caim_multifile
• Fixed issue #844
• Docs updates

1.8.6

• Remove some blockers for filepath support in the codebase
• Several bugfixes
• Early performance guide
• Some support for loading videos from MKV files
• Docs improvements

See #837 for a more detailed list of changes

1.8.5

• New spike selection method
• Some speed/memory improvements
• Code cleanups
• restore estimates.W as csr_matrix when loading object saved as hdf5
• Some dependency unpinning
• fix saveNWB for 1p data
• Support minimum spike size (s_min) for p=2
• Misc bugfixes

1.8.4

Fixing naming of release for conda tooling

v1.8.3

*Compatibility with tensorflow 2
*Compatibility with bokeh2
*3d demo fixes
*More options for AVI movie saving
*recompute estimates.coordinates when running refit

1.8.2

• GUI: Fix some visualisations
• Motion correction: Improved validation for setup, to catch rare issues
• Fix some incorrect parameters in the memmap code
• CNMF: Some improved defaults
• Jupyter Demos: Fix some issues relating to tighter validation upstream in notebook formatting
• Packaging: Add psutil dependency explicitly in case conda environment does not provide it

1.8.1

• Remove keras from list of dependencies to avoid problems arising sometimes in windows installations #702 This does not affect the codebase unless you want to use the VolPy module for analyzing voltage imaging data. In this case, you would need to install keras by entering your environment and typing
  conda install keras -c conda-forge
• Turn the GUI into binary and make it accessible through the caiman_data folder #700
• Fix a bug occurring sometimes when plotting contours in a notebook #699