- Improvements:
- Automatic conversion to
int
using the function signature will now use the type annotations, then the default value if no annotation is present.Optional[int]
values will be converted toint
.
- Automatic conversion to
- Bug fixes:
- In some distributions of MATLAB the automatic conversion from Numpy
ndarray
using MATLAB'sdouble
does not work. If it fails, convert to a regularpy.array
first, this should be more reliable.
- In some distributions of MATLAB the automatic conversion from Numpy
- Bug fixes
- Check for correct Matlab version (
>=2019a
) when converting Numpy arrays withdouble
. Converting withdouble
works for Numpy arrays with>=2019a
, but only works for native py.array.array arrays in2018a/b
.
- Check for correct Matlab version (
- Improvements
- Only warn once if converting a Numpy array in an old Matlab version (less than
2019a
) - Only determine Matlab version once for increased performance
- Only warn once if converting a Numpy array in an old Matlab version (less than
This release changes the API - now in addition to the pyobj property, subclasses must declare a classname
which is a string being the Python class name of the wrapped class, obtained from type(obj)
in Python.
This change enables the new overloaded help
and doc
functions to obtain a reference to the Python objects to get docstrings using the pydoc
system. In addition it means that users no longer have to define the helpobj property as this can now be obtained from the classname.
Adds a static method to direct Python warnings to sys.stdout which is mirrored by Matlab, unlike sys.stderr.
First alpha release.
- Allows Matlab to wrap arbitrary Python classes.
- Allows saving Python classes to mat files (data is pickled so any classes which cannot be pickled will not be saved correctly).
- Overloads help command to print pydoc help text in Matlab.