.. _release0.94.0:

SpikeInterface 0.94.0 release notes
-----------------------------------

15th April 2022


Main change:

 * Refactor WaveformExtractor with waveform_tools.
 * Implement Zarr backend for save()
 * Read IBL compressed files.
 * Phase shift (destripe) preprocessor
 * Test are run partially : faster GH actions
 * Many improvement in sorting compnents: template matching, select_peaks, motion_estimation, motion_correction


core:

 * Fix typo in SplitSegmentSorting init
 * Added NumpySorting.from_peaks()
 * Refactor WaveformExtractor with waveform_tools.
 * Fix bug "too many open files" for the new waveforme extractor.
 * Add function to remove empty units
 * Remove loose version in favor of packaging
 * Add chunk_duration concept for jobs
 * Implement Zarr backend for save()
 * Add split recording class and function
 * Add Recording.remove_channels()
 * Add save sorting to memory



extractors:

 * Reintroduce spikeglx test files
 * Add NeuroscopeSortingExtractor
 * New neo extractors
 * Add more neualynx versions for testing
 * Modify nwb recorder extractor for loading channel name property as channel_ids when available
 * Support 3d positions (rel_z) in NWB
 * Read openephys probe from probeinterface
 * Fix "KSLabel" extraction in phykilosortextractor
 * Bug fix on waveclus sorting extractor
 * Read IBL compressed files



sorters:

 * Refactor run_sorters/run_sorter_by_group to better handle folders and key dtypes

toolkit:

 * Phase shift (destripe) preprocessor
 * Fix typo in isi violation
 * Addresses batch size bug #565, and also resolves open file warnings
 * Add nan to d_prime for single unit case
 * Handle computing of nearest neighbor metrics for units with few spikes


comparison:

 * Fix multisortingcomparison spiketrains

widgets:

 * Add presence plot
 * Add with_channel_ids for several widgets that use plot_probe
 * Add optional distinctipy


export:

  * Fix mismatch between phy and si unit ids


sortingcomponents:

 * Improvements to template matching
 * Start sparse clustering
 * Improvements to select_peak()
 * Improvements to motion_estimation()
 * Improvements to motion_correction()



doc:
 * Documentation updates for quality_metrics
 * Try install SI from conda env and skip pip
 * Improve documentation for sorting components

 CI:

  * Refactor tests
  * Partial tests
  * Fix herdingspikes version for tests