PyCall.jl

Fixes type instability (#672), aot compilation (#651, #673), pyraise (#675).

Fix #670.

Various fixes (#661, #668), add float(::PyObject) (#660), extend Base.hasproperty in Julia 1.2.

In preparation for PyCall 2.0 (#629), this deprecates the old o[:foo] access in favor of o.foo (#517).

Old PyCall-using packages should switch over to the new style in order to eliminate zillions of deprecation warnings, and put PyCall 1.90 in their REQUIRE file.

More sophisticated libpython location (#556) and various bugfixes (#579, #558, #580, #571)

Bugfix (#553).

Bugfix for Python 3.7 with Date conversions (#540).

Bugfix for #534.

Fix #533.

1.0 compatibility (#516, #522), default to python3 on Unix (#512), pycall! function (#492).

More 0.7 deprecation fixes.

More 0.7 fixes, makes pybytes constructor work for any stride-1 DenseVector{UInt8}.

bugfix for handling of NumPy scalars in Python 3

Adds pyfunctionret, ispynull, pyimport_e, uses the VersionParsing package for parsing the Python version number, fixes a couple of bugs, drops Julia 0.5 support, and improves 0.7 support.

Many pyjulia fixes. Also supports iterating over Julia iterators from Python.

More flexible tuple conversions from PyObect, fixes to I/O wrappers to make them conform to the Python docs (and not strip newlines in 0.6).

Added @pywith (https://github.com/JuliaPy/PyCall.jl/pull/375), pybytes, serialization via pickle (https://github.com/JuliaPy/PyCall.jl/pull/388). Made dictionary conversion more conservative (https://github.com/JuliaPy/PyCall.jl/pull/384).

This release adds support for binary operators like + on Python objects and (in 0.6) in-place updating operators like .+= (mapping to Python __iadd__ methods), it generalizes the append! function to work on non-list objects, and fixes some build problems on Windows. 0.4 support is dropped in this release.

This release adds support for binary operators like + on Python objects and (in 0.6) in-place updating operators like .+= (mapping to Python __iadd__ methods), it generalizes the append! function to work on non-list objects, and fixes some build problems on Windows. 0.4 support is dropped in this release.

This fixes another deprecation warning (JuliaPy/PyPlot.jl#283) and improves robustness against user PYTHONPATH environment variables when using Conda (JuliaPy/PyPlot#286).