Version 0.8.1

This is a minor update with the following changes and additions:

• Added shortcuts for common queries with pandas accessors (#133)
• Deferred path finding until necessary (#135)
• Significantly improved read speed for split EEG data (#137)

Version 0.8.0

• Added support for loading pyFR data (#117)
• Simplified EEG loading by removing the option to load directly via epochs
  (#125)
• Renamed the class holding results from CMLReader.load_eeg to
  EEGContainer to avoid confusion with the PTSA TimeSeries class (#126)
• Added a new CMLReader.load_events classmethod to load events from
  multiple subjects and/or experiments (#129)
• Added support for loading multisession EEG data (#130)

Version 0.7.2

• Improved conversion of EEG data to PTSA format (#107)
• Fixed loading events for PS2 and PS4 (#110, #112)
• Improved error message when trying to load EEG with an empty events DataFrame
  (#114)

Version 0.7.1

New feature:

• Results of get_data_index are now cached using functools.lru_cache
  (#101).

Bug fix:

• Magic importing of reader classes didn't work if not in a specific working
  directory (#104). Fixed in PR #105.

Version 0.7.0

User-facing changes:

• Localization and montage numbers are now converted to integers instead of
  being strings (#91)
• Fixed loading of montage data for subjects with a montage number other than
  0 (#95)
• Added preliminary support for loading ltp data (#97)

Other changes:

• CI testing updated to use an environment variable to specify what Python
  version to run (#93)
• Test data gets written to a temporary directory instead of polluting the
  cmlreaders.test.data package (#96)
• Reader classes are automatically discovered instead of having to specify them
  in cmlreaders/readers/__init__.py (#99)

Version 0.6.0

This release fixes several bugs with EEG reading when passing a referencing
scheme and improves performance when loading pairs/contacts data. Highlights:

• Adds and improves existing test cases for rereferencing EEG data
• Improved load speed of pairs.json/contacts.json by about 2 orders of
  magnitude (#89)
• Speeds up loading of split EEG data when specifying a referencing scheme by
  only loading the required data (#85)

v0.5.0

New features:

• Allow globally setting the root directory with an environment variable (#46)
• Added a function to check if EEG data can be rereferenced
• Automatically determine montage and localization numbers when possible (#77)
• Added a fromfile method to classes based on BaseCMLReader to more
  easily directly load specific data types (#79)

Improvements:

• Added support for reading EEG data from restarted sessions (#68)
• Improved the ergonomics of passing a scheme keyword argument to
  rereference EEG data (#70)
• Make channel filtering via the scheme keyword argument more explicit (#80)

Bug fixes:

• Handle loading PS4 events (#47)
• Fixed paths with respect to montage/localization confusion (#62)
• Fixed the CSVReader to correctly read jacksheets (#65)
• Handle gaps in contact numbers when reading EEG data (#63)

v0.4.0

Summary of Changes:

• Implemented custom TimeSeries representation that can be converted to PTSA or MNE format
• Implemented EEG reader with support for loading a full session or event-based subset
• Updated getting started guide and documentation

v0.3.1

Minor bugfix to setup.py that was preventing the conda package from building correctly

v0.3.0

This is a relatively large release as it modifies the API from 0.2.0 and implements readers for most of the RAM data, except for EEG.