What’s new¶
version 1.7.4¶
Not yet released
- Changed behaviour of Constructs.filter_by_axis.
- New methods: Data.has_units, Data.has_calendar, Data.has_fill_value.
- New keyword ‘constructs’ to Field.transpose.
- Keyword ‘axes’ to Field.set_data is now optional.
- Added the ‘has_bounds’ method to constructs that have data but can’t have bounds.
- New methods: DomainAxis.nc_is_unlimited, DomainAxis.nc_set_unlimited.
- Made Data a virtual subclass of Array.
- Deprecated methods: Field.nc_unlimited, Field.nc_clear_unlimited, Field.nc_clear_unlimited.
- Fixed bug when writing new horizontal coordinate reference for the vertical datum.
- Fixed bug in del_data method.
- Fixed bug with in-place operations.
- Fixed bug with position in some insert_dimension methods.
- Fixed bug that sometimes made duplicate netCDF dimensions when writing to a file.
- Added _shape keyword to Field.set_data_axes to allow the data shape to be checked prior to insertion.
- Added the ‘_custom’ attribute to facilitate subclassing.
- New class mixin.NetCDFUnlimitedDimension replaces mixin.NetCDFUnlimitedDimensions, which is deprecated.
- New method CFDMImplementation.nc_is_unlimited_axis replaces CFDMImplementation.nc_get_unlimited_axes, which is deprecated.
- New method CFDMImplementation.nc_set_unlimited_axis replaces CFDMImplementation.nc_set_unlimited_dimensions, which is deprecated.
version 1.7.3¶
2019-04-24
- New method: Constructs.filter_by_size
- New method: Data.uncompress
- Changed the default behaviours of the Construct.filter_by_axis, Construct.filter_by_size, Construct.filter_by_naxes, Construct.filter_by_property, Construct.filter_by_ncvar, Construct.filter_by_ncdim, Construct.filter_by_method, Construct.filter_by_measure methods in the case when no arguments are provided: Now returns all possible constructs that could have the feature, with any values.
- Renamed the “underlying_array” methods to “source”
- Added _field_data_axes attribute to Constructs instances.
- Added _units and _fill_value arguments to get_data method.
- Moved contents of cfdm/read_write/constants.py to NetCDFRead and NetCDFWrite.
- Fixed bug in CoordinateReference.clear_coordinates
- Fixed bug in Field.convert (which omitted domain ancillaries in the result)
- Added **kwargs parameter to CFDMImplementation.initialise_Data, to facilitate subclassing.
- Added NetCDFRead._customize_read_vars to facilitate sublcassing.
- Added NetCDFWrite._transform_strings to facilitate sublcassing.
version 1.7.2¶
2019-04-05
- New “mode” parameter options to Constructs.filter_by_axis: ‘exact’, ‘subset’, ‘superset’
- Enabled setting of HDF5 chunksizes
- Fixed bug that caused coordinate bounds to be not sliced during subspacing.
version 1.7.1¶
2019-04-02
- New methods Constructs.clear_filters_applied, Constructs.filter_by_naxes
- Changed behaviour of Constructs.unfilter and Constructs.inverse_filters: added depth keyword and changed default