Added
- Initial public release.
- SPP (Single-Point Positioning) solver with weighted least-squares.
- RINEX observation (`.obs`) and navigation (`.nav`) readers.
- DJI `.MRK` ground-truth waypoint reader.
- Simple CSV readers/writers for observations, trajectories, and waypoints.
- Trajectory resampling by fixed time interval or fixed distance.
- Waypoint-based trajectory optimisation (ECEF piecewise-linear residual field).
- CZML and GeoJSON export for Cesium visualization.
- TOML / JSON / INI configuration loader and validator.
- End-to-end pipeline orchestrator and CLI entry point.
- Synthetic GNSS constellation for self-contained demos and tests.
- Bilingual README (中文 / English) with architecture diagrams.
- MIT license.
Fixed
- `examples/run_full_pipeline.py` now correctly reads MRK waypoints from CSV
(previously used the DJI `.MRK` reader on a CSV file, yielding 0 waypoints
and no optimisation effect). The end-to-end demo now shows the expected
5.58 m → 0.85 m RMS error reduction.
- `csv_io.read_waypoints_csv` now accepts common header aliases
(`time`/`t`, `latitude`/`lat`, `longitude`/`lon`, `ellh`/`height`).
- `examples/run_full_pipeline.py` loads the sibling `generate_sample_data.py`
robustly via `importlib` instead of a fragile relative import.
- Removed a hard-coded local absolute path from `examples/run_dji_real.py`;
the script now requires the data directory to be passed explicitly.
Changed
- `pyproject.toml` now includes full PyPI metadata (license, authors,
keywords, classifiers, URLs, optional dev dependencies).
- Added `.gitignore` to exclude caches, build artifacts, and generated examples.