CHANGELOG.md

[UNRELEASED] - YYYY-MM-DD

Fixed

• Expose detected segment (used in dejittering) as stream["info"]["segments"] (#117 by Robert Guggenberger)
• A non-looping playback of an XDF file will no longer lead to a TypeError (#119 by Stefan Appelhoff)
• Reported segments now have the correct sample stop, are no longer duplicated, and empty streams return segments as an empty array rather than [0, -1] (#121 by Jamie Forth)
• Test unduplicated reporting for segments (#123 by Robert Guggenberger, Clemens Brunner)

Changed

• Rename pyxdf.examples module to pyxdf.cli (#118 by Clemens Brunner)
• Reverse logic of --loop argument in pyxdf.examples.playback_lsl.py to be more in line with standard practice: Supplying --loop will loop, whereas omitting --loop will NOT loop (#119 by Stefan Appelhoff)

[1.16.8] - 2024-07-18

Fixed

• Fix reading numeric array data on big-endian hosts (#98 by Ben Beasley)

[1.16.7] - 2024-07-17

Added

• Add clock_times and clock_values to stream dictionary (#105 by Fabian Grosch)

[1.16.6] - 2024-04-18

Changed

• Minimum required Python version is now 3.9 (#100 by Clemens Brunner)

[1.16.5] - 2024-01-12

Added

• Added compatibility with Python 3.12 (#96 by Clemens Brunner)

[1.16.4] - 2023-01-09

Fixed

• Fix invalid np.object type (#87 by Clemens Brunner)
• Fix robust fit for large timestamps (#73 by Chadwick Boulay)
• Fix loading stream with no samples when dejitter_timestamps=False (#82 by Robert Guggenberger)

[1.16.3] - 2020-08-07

Added

• Add Cython type hints (requires optional local compilation) (#17 by Tristan Stenner)

Fixed

• Handle XDF files with corrupt chunk headers (missing stream IDs) more gracefully (#62 by Robert Guggenberger)

• Treat nominal_srate field as float to fix parsing errors (#65 by Robert Guggenberger and #68 by Clemens Brunner)

Changed

• load_xdf now requires keyword-only arguments after the first two arguments (#59 by Christian Kothe)

[1.16.2] - 2019-10-23

Added

• Allow loading from already opened file objects, e.g. in-memory files or network streams (#51 by Tristan Stenner)
• Add CI tests with example data (#49 by Clemens Brunner)

Fixed

• Compare nominal to effective sampling rates only for regularly sampled streams (#47 by Clemens Brunner)
• More robust error recovery for compressed corrupted files (#50 by Tristan Stenner)

Changed

• Speed up loading of numerical data (#46 by Tristan Stenner)
• Avoid/suppress some NumPy warnings (#48 by Clemens Brunner)

[1.16.1] - 2019-09-28

Fixed

• Remove Python 2 compatibility from setup.py (#45 by Clemens Brunner)

[1.16.0] - 2019-09-27

Added

• Add option to load only specific streams (#24 by Clemens Brunner)
• Add verbose argument to load_xdf (#42 by Chadwick Boulay)

Fixed

• Fix bug in jitter removal (#35 by Alessandro D'Amico)
• Add compatibility with Python 3.5 by converting Pathlike objects to str for file open functions (#37 by hankso)

Changed

• Refactor jitter removal code to be more readable (#36 by Chadwick Boulay)

[1.15.2] - 2019-06-07

Added

• Store unique stream ID inside the ["info"]["stream_id"] dict value (#19 by Clemens Brunner)

[1.15.1] - 2019-04-26

Added

• Support pathlib objects (#7 by Clemens Brunner)
• Allow example script to be called with an optional XDF file name (e.g. python -m pyxdf.example or python -m pyxdf.example xdf_file.xdf) (#10 by Tristan Stenner)

Fixed

• Use correct data types (#2 by Tristan Stenner)
• Streams are not sorted by name anymore (#3 by Clemens Brunner)
• Support older NumPy versions < 1.14 (#4 by Clemens Brunner)
• Pass correct on_chunk argument (#5 by Clemens Brunner)
• Fix _scan_forward (#6 by Clemens Brunner)

Changed

• Pull StreamData class and chunk 3 reader out of load_xdf (#13 by Tristan Stenner)