- First release on PyPI.
- Fixed bug in setup.py version number
- pre-commit.ci bug fixed
- Read version number from __init__
- Expand HISTORY.rst
- Outsource useful functions for reading and writing large netCDF files.
- Use those functiosn directly from pyhomogenize.
- Calling pyhomogenize classes is not needed.
- rename save_to_netcdf to save_xrdataset
- write input files to ds attributes
- create chunks if not already existing
- write CF variables to dataset while calling open_xrdataset
- add data via pip install
- precise mid of time range
- add more dependencies (cftime, netcdf4, h5netcdf)
- add decode_times and combine as keyword argumnets to open_xrdataset
- set default value of parallel in open_xrdataset to False
- save_xrdataset is now working if Dataset does not include time
- convert time axis to CFTimeIndex
- set proleptic_gregorian to standard calendar
- calculate time_bnds and add them to dataset
- get CF_variables
- read and write data with length of time axis is 1
- add time bounds with length of time axis is 1
- drop coordinates from non CF variables
- testing for python version 3.8, 3.9 and 3.10
- published on zenodo
- add time bounds if length of time axis is 1
- convert np.datetime64 object to CFTimeIndex
- fixing v0.5.1
- convert time axis to cftime.datetime object
- open_xrdataset: set decode_cf to parameters
- some restructuring in _convert_time
- use unlimied_dims only if dimension name in ds.coords when saving dataset as netcdf file
- remove Sphinx from requirements_dev.txt
- new frequency time2: YY1-07-01 - YY2-06-30
- convert time_ranges to CFTimeIndex
- make time_control more flexible
- new frequency: yearAC (yyyy1-07-01 to yyyy2-06-30)
- new_frequency: yearHydro (yyyy1-09-01 to yyyy2-08-31)
- add time bounds for daily data
- testing python versions 3.9 to 3.11