Skip to content

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

The same types of changes should be grouped. Types of changes:

  • Added for new features.
  • Changed for changes in existing functionality.
  • Deprecated for soon-to-be removed features.
  • Removed for now removed features.
  • Fixed for any bug fixes.
  • Security in case of vulnerabilities.

[Unreleased]

[0.30.0] - 2026-05-04

Changed

  • WorkoutData.resample(freq, interpolate=False) semantic downsampling behavior for lower-frequency buckets: cumulative fields (distance) use max; instantaneous fields (speed, power, heartrate, cadence, temperature, grade, elevation, enhanced_altitude) use mean; GPS coordinates (latitude, longitude) use mean; is_moving uses any. Accepts any pandas offset alias (e.g. "10s", "1min").
  • Development tooling migrated from Poetry/Black to uv/Ruff: local and container workflows now use uv sync/uv run, linting uses ruff format and ruff check, and Poetry-specific artifacts/config (including poetry.lock) were removed.
  • Contributor and project docs were updated for the uv workflow (CONTRIBUTING.md, README.md, docs landing page, and lab notebook setup instructions), and project automation guidance (Makefile, agent/copilot instructions) now reflects uv + ruff commands.
  • Workspace defaults now pin Python 3.13 via .python-version and point VS Code to .venv as the default interpreter.

Fixed

  • Fixed ValueError in elevation_smooth_time() for short recordings where the data length is less than the Savitzky-Golay filter's default window length (21). The function now gracefully degrades: recordings too short for any valid window are returned unsmoothed; recordings shorter than the default window use a reduced (but still valid) odd window length.

[0.29.1] - 2026-04-29

Changed

  • Python version support: Dropped Python 3.10, added Python 3.11, 3.12, 3.13, and 3.14 to test matrix
  • Dependency upgrades: Updated all dependencies to current stable versions
  • numpy: >=1.21,<1.24 → ^1.24.0 (supports numpy 2.x)
  • pandas: ^1.5.2 → ^2.0.0 (pandas 2.x with improved performance and new features)
  • scipy: ^1.4.1 → ^1.13.0
  • scikit-learn: >= 0.23.1 → ^1.3.0
  • lmfit: ^1.0.0 → ^1.3.0
  • fitparse: ^1.1.0 → ^1.2.0
  • Dependency restructure: matplotlib, requests, and mkdocs-material removed from main dependencies (were never part of the public API). dev and docs dependency groups are now optional — install with poetry install --with dev,docs for local development.

Fixed

  • Fixed resample_data() compatibility with pandas 2.x by adding numeric_only=True to mean() aggregation
  • Fixed tox configuration inconsistency (now tests py311, py312, py313, py314 instead of py39, py310)
  • Migrated pydantic v1 APIs to v2: @validator → @field_validator, parse_obj_as → TypeAdapter.validate_python, parse_obj → model_validate
  • Replaced deprecated scipy.interpolate.interp1d with numpy.interp (linear) and scipy.interpolate.make_interp_spline (quadratic) in elevation smoothing
  • Fixed SettingWithCopyWarning in TCX parser by calling .copy() after dropna()
  • Fixed deprecated pd.to_numeric(errors="ignore") in TCX parser
  • Fixed deprecated XML element truth-value test in TCX parser (if extensions: → if extensions is not None:)

[0.29.0] - 2026-04-04

Added

  • WorkoutData.merge_many() class method to merge a list of separately recorded workouts into a single continuous WorkoutData object. Workouts are sorted by start timestamp; overlapping timestamps favour the later workout; time gaps are preserved as NaN rows after resampling (when interpolate=False). Accepts drop_gaps=True to shift workouts back-to-back (no gap rows), which condenses elapsed time — note that performance metrics will not account for rest time in this mode.
  • WorkoutData.merge() instance method as a convenience wrapper around merge_many(), also accepts drop_gaps.
  • distance is forward-filled across gap rows in merged workouts (it is cumulative and should not appear as NaN during rest periods).
  • WorkoutData.set_start_time() method to shift the datetime index of a workout to a specified start time, recomputing the time column. Useful when a file (e.g. a Strava activity stream) contains no absolute timestamp.
  • Three new Strava activity-stream example files covering a single session split across three recordings: strava_17497731832_warm_up.json, strava_17497955000_track_workout.json, strava_17498168651_warm_down.json.
  • Added lab/merge_workouts.ipynb notebook demonstrating merge_many(), merge(), and set_start_time() with the new example files.

Changed

  • Updated pyproject.toml build backend from poetry.masonry.api to poetry.core.masonry.api to support editable installs (pip install -e .).

[0.28.2] - 2026-04-03

Added

  • Added enhanced_speed and enhanced_altitude to DataTypeEnum to support newer Garmin FIT files that use enhanced fields.
  • Added docs/Example.ipynb notebook demonstrating the Osaka Marathon 2025 dataset, included in the docs site navigation.
  • Added Releasing section to CONTRIBUTING.md documenting the full release workflow.
  • Created lab/ directory with getting_started.ipynb (core API walkthrough) and osaka_marathon.ipynb (real-world running example using the Osaka Marathon 2025 FIT file).
  • Documented lab/ directory setup in CONTRIBUTING.md for faster local development and prototyping — useful when forking or onboarding to the project.

Changed

  • WorkoutData.from_file() now accepts an ExampleData object directly (no need to pass .path explicitly).
  • Updated examples/index.yml for the Osaka Marathon FIT file to use the correct column names (enhanced_speed, enhanced_altitude).

[0.28.1] - 2025-05-23

Changed

  • Changed test_core tests to handle changed best_interval and multiple_best_intervals.
  • Updated homepage and documentation URLs in pyproject.toml.
  • Improved and clarified documentation.

Fixed

  • Fixed linting to conform with black so lint.yml passes.
  • Removed deprecated poetry.dev-dependencies and updated poetry.lock.
  • PyPI homepage and documentation URLs.

[0.28.0] - 2025-05-19

Added

  • Extended WorkoutData with new methods: best_intervals, best_distance_intervals, fastest_distance_intervals, and elevation_gain for easier access to interval and elevation metrics.
  • Added fastest_distance_interval and multiple_fastest_distance_intervals functions for finding the fastest intervals over fixed distances.
  • Added new elevation-related functions: elevation_gain, elevation_smooth_time, and elevation_smooth for calculating and smoothing elevation data.
  • Added new grade-related functions: grade_smooth_time, grade_smooth, and grade_raw for calculating smoothed and raw grade (slope) values.
  • WorkoutData now automatically computes a new time column containing the relative time (in seconds) since the workout started, in addition to the datetime index.
  • WorkoutData now automatically adds an is_moving column based on the speed column.
  • WorkoutData now automatically computes and adds a grade column if both distance and elevation are present.

Changed

  • best_interval and multiple_best_intervals now return a dictionary with the max mean value and the start/stop indices of the best interval for a provided time window or list of time windows, for consistency with distance-based interval functions.

[0.27.0] - 2025-05-15

Added

  • best_distance_interval() function for finding the fastest interval over a fixed distance.
  • WorkoutData, a pandas.DataFrame-like class for standardising the workout object structure from different file types or sources.

Changed

  • Refactored changelog to adopt Keep a Changelog format fully.

[0.26.1] - 2025-05-09

Added

  • read_local_strava() function for loading locally saved Strava activity stream data in JSON format.
  • Example file added at examples/data/ to demonstrate usage of read_local_strava().

Changed

  • Project forked from sweatpy, continuing from version 0.25.0.
  • Project renamed from sweatpy to chironpy, including updates to all import paths, README, and documentation.
  • Added Python 3.11, updated dependencies.

[0.25.0] - 2022-02-01

Added

  • Added full support for parsing fit, tcx and gpx courses. Some of the functionality was already working, this release makes it complete.

[0.24.0] - 2021-12-16

Added

  • Adds processing of resting heartrate to athlete model from FIT files.

Fixed

  • Adds fixes for processing age and heigh to athlete model from FIT files.

[0.23.2] - 2021-12-13

Fixed

  • Fixes parsing the activity_class attribute on the athlete model returned by read_fit(metadata=True). It should now handle all possible values.

[0.23.1] - 2021-11-11

Fixed

  • Some FIT files contain records with duplicate field names of which only 1 was not null. Added code to make sure that the non null value was picked.

[0.23.0] - 2021-11-01

Changed

  • Updated dependency versions. Most notable is scikit-learn to ">= 0.23.1".

Deprecated

  • Python 3.6 support will be dropped in a next release as it is nearing EOL and some dependencies like scikit-learn already do not support it in their latest releases.

[0.22.1] - 2021-10-19

Fixed

  • Relax Athlete model validation because it raised errors for some FIT files.

[0.22.0] - 2021-10-19

Added

  • When metadata=True is passed to chironpy.read_fit() the response now includes an "athlete" key that contains an Athlete model with information about the athlete.

Fixed

  • Suunto hrv samples that consist of only 1 value are now properly handled and do not raise a TypeError anymore.

[0.21.0] - 2021-04-22

Added

  • Adds testing for Python 3.10
  • read_tcx() now supports a metadata=True kwarg that will return device data.

[0.20.3] - 2021-04-20

Fixed

  • chironpy.read_fit() now supports integer column names.

[0.20.2] - 2021-04-16

Fixed

  • read_fit() now handles summary data correctly when there is no data

[0.20.1] - 2021-04-16

Fixed

  • Fixed issue when pool_lengths=True but there are no pool length records in the FIT file.

[0.20.0] - 2021-04-15

Added

  • read_fit() now returns pool length records when thepool_lenghts=True` argument is passed.
  • read_fit() now return raw FIT messages when a raw_message=True argument is passed.
  • read_fit() can now return metadata and device information.

[0.19.0] - 2021-04-06

Added

  • Adds matplotlib dependency.
  • Extends chironpy.read_fit() with support for multi session FIT files, hrv data, metadata (summaries and devices), error handling and much more...
  • Properly handles left-right balance from FIT files.
  • Adds Garmin FIT Profile based on the Garmin FIT SDK 21.47.

Fixed

  • Removes unnecessary numpy import in pdm docs.

[0.18.1] - 2021-03-12

Fixed

  • Fixes the index of mean max calculations as they were of by 1 second. First index is now 00:00:01 instead of 00:00:00.

[0.18.0] - 2021-03-12

Added

  • Adds monotonic argument to chironpy.mean_max() to enforce a monotonically decreasing mean max curve. Usage: power.chironpy.mean_max(monotonic=True). Defaults to False.

[0.17.1] - 2021-03-09

Fixed

  • Numerical columns are now converted to numeric dtypes for gpx files.

[0.17.0] - 2020-12-31

Fixed

  • Fixes error when loading FIT file without location data.

Changed

  • Removes "forward fill" when resampling dataframes. Users can manually get the old behavior back by running df = df.ffill().

[0.16.1] - 2020-11-19

Fixed

  • Fixes publishing docs by updating mknotebooks version >0.6.0

[0.16.0] - 2020-11-19

Added

  • Adds support for Python 3.9

Fixed

  • Latitude and longitude from FIT files are now converted from semicircles to degrees.

[0.15.0] - 2020-06-14

Added

  • Adds exponential power duration model.
  • Adds omni power duration model

Changed

  • Renames CriticalPowerRegressor to PowerDurationRegressor

[0.14.0] - 2020-06-09

Added

  • Adds chiron accessor to pandas series and data frames.
  • Add to_timedelta_index() and mean_max() to accessors.
  • Adds calculate_zones() and time_in_zone() to series accessor.

[0.13.0] - 2020-06-05

Added

  • Rewrite of Strava implemention, now features a single read_strava() function.

Removed

  • Old Strava functionality is removed in favor of a new read_strava() function.

[0.12.0] - 2020-06-05

Added

  • Added the CriticalPowerRegressor
  • Some refactoring and documentation for W'balance functionality.

Removed

  • Removed the old critical power curve fitting functionality
  • Removed WorkoutDataFrame model because it is not maintained.
  • Removed Athlete model because it is not maintained.
  • Removed chironappClient because it is not maintained. Will be re-implemented later.

[0.11.0] - 2020-06-04

Added

  • Adds lap and session data for FIT files
  • Adds lap data for TCX files

[0.10.0] - 2020-06-01

Added

  • Usage examples on docs home page
  • Nomenclature for data frames
  • Adds ability to resample and interpolate data frames

Changed

  • Changes the index of returned data frames to a pandas.DatetimeIndex

[0.9.0] - 2020-05-23

Added

  • Added generic read_file function
  • Added generic read_dir function

[0.8.0] - 2020-04-26

Added

  • Added reading TCX files.

[0.7.0] - 2020-04-21

Added

  • Added reading GPX files.

[0.6.0] - 2020-04-21

Added

  • Added example data framework

Changed

  • Refactored reading FIT files to pandas-like interface: read_fit()

[0.5.0] - 2020-17-04

Added

  • Python 3.8 support.
  • Tests are now run in docker.
  • Implements Poetry for packaging and dependencies.
  • Adds Github Actions for tests and publishing.
  • Updates copyright year in LICENSE.
  • Adds CONTRIBUTING.md.
  • Adds Jupyter notebook functionality for docs.