## PythonCall v0.9.31 [Diff since v0.9.30](https://github.com/JuliaPy/PythonCall.jl/compare/v0.9.30...v0.9.31) **Merged pull requests:** - Bump actions/checkout from 5 to 6 (#726) (@dependabot[bot]) - Bump actions/cache from 4 to 5 (#732) (@dependabot[bot]) - support python 3.14 (#733) (@cjdoris) **Closed issues:** - dtype of NumPy array created from Julia array is `object` on Julia nightly (#439) - Intermittent CI failures: Fatal Python error: _Py_CheckRecursiveCall: Unrecoverable stack overflow (#694) - using PythonCall in Pluto crashes Pluto Malt.TerminatedWorkerException (#711) - Julia crashes (Segmentation fault) when using tab completion with PythonCall (#713) - Make `ArrayValue` not callable (#728) - ImportError when using dolfinx in PythonCall (#729) - Segfault when calling Python function repeatedly. (#731)

## PythonCall v0.9.27 [Diff since v0.9.26](https://github.com/JuliaPy/PythonCall.jl/compare/v0.9.26...v0.9.27) **Merged pull requests:** - Test with PYTHON_JULIACALL_HANDLE_SIGNALS=yes (#642) (@cjdoris) - Switch to heap types using PyType_FromSpec (#643) (@cjdoris) - Bump minimum supported Python version from 3.8 to 3.9 (#645) (@cjdoris) - Remove dead code supporting old versions of Julia/Python (#647) (@cjdoris) - Centralize Public API into dedicated API module (#648) (@cjdoris) - Bump actions/checkout from 4 to 5 (#652) (@dependabot[bot]) - build(deps): add libstdcxx pin, bump CondaPkg (#656) (@cjdoris) **Closed issues:** - Switch to heap types (#641) - SSL CERTIFICATE_VERIFY_FAILED (#651)

## PythonCall v0.9.26 [Diff since v0.9.25](https://github.com/JuliaPy/PythonCall.jl/compare/v0.9.25...v0.9.26) **Merged pull requests:** - bump min required julia version to 1.9 (#614) (@cjdoris) - use getptr a lot less (#618) (@cjdoris) - use uv more (#621) (@cjdoris) - Bump astral-sh/setup-uv from 5 to 6 (#623) (@dependabot[bot]) - docs: Code loading order (#625) (@icweaver) - Improve GitHub workflow job names for better clarity (#633) (@cjdoris) - Add PySide6 support to gui.jl (#634) (@cjdoris) - CI benchmarking suite take 2 (#636) (@MilesCranmer) - Add FAQ on interactive threads (#638) (@visr) - hotfix: misaligned pointer cast (#640) (@MilesCranmer) **Closed issues:** - Segmentation fault when calling basic python functions (#617) - Segmentation fault in basic example from docs (#624) - Segfault when calling python function repeatedly (#626) - `pyimport` can change `set_zero_subnormals` (#629) - Support PySide6 in Compat (#632)

## PythonCall v0.9.25 [Diff since v0.9.24](https://github.com/JuliaPy/PythonCall.jl/compare/v0.9.24...v0.9.25) **Merged pull requests:** - Add `--heap-size-hint` parameter (#547) (@MilesCranmer) - Add `elsize` method for PyArray (#593) (@cjdoris) - Bump actions/cache from 1 to 4 (#602) (@dependabot[bot]) - Bump julia-actions/setup-julia from 1 to 2 (#603) (@dependabot[bot]) - Bump actions/checkout from 2 to 4 (#604) (@dependabot[bot]) - Bump codecov/codecov-action from 1 to 5 (#605) (@dependabot[bot]) - Bump actions/setup-python from 3 to 5 (#606) (@dependabot[bot]) - OpenSSL compat (#613) (@cjdoris) **Closed issues:** - Specifying `--heap-size-hint` to JuliaCall (#546) - Error for code with PythonCall that works with PyCall (#576) - `Base.elsize` not implemented for `PyArray` (#579) - pytorch GC (#592) - Precompile cache misses (#594) - PythonCall not working in Jupyter Notebook in VSCode (#601)

## PythonCall v0.9.24 [Diff since v0.9.23](https://github.com/JuliaPy/PythonCall.jl/compare/v0.9.23...v0.9.24) **Merged pull requests:** - Switch to pyproject.toml (#545) (@MilesCranmer) - fix: patch changed behavior of `setproperty!` for modules (#583) (@MilesCranmer) - PyList: fix pushfirst! for Julia 1.11, by adding prepend! (#588) (@hhaensel) **Closed issues:** - Segmentation fault when `juliacall` is imported before `torch` (#435) - `pyconvert_add_rule` must be executed before using `pyimport` in `__init__()` (#500) - juliacall 0.9.22 can result in hanging precompilation (#537) - Segmentation Fault in combination with Distributed (#548) - Segfault with package, but not with identical module (#554) - OpenSSL library collision with external python on linux systems (#561) - How to debug `openssl` could not resolve for environment spec inside a package? (#562) - `no method matching project_rel_path(::Pkg.Types.EnvCache, ::Nothing)` on 1.10.5 (#564) - General registry segfault (#569) - Broken on Python 3.13.0 (#581) - Julia 1.11 prevents assigning objects to Main (#582)

## PythonCall v0.9.22 [Diff since v0.9.21](https://github.com/JuliaPy/PythonCall.jl/compare/v0.9.21...v0.9.22) **Merged pull requests:** - Require a compatible openssl version (#523) (@lassepe) - More thread-safe GC (#529) (@cjdoris) - simplify GIL handling (#530) (@cjdoris) - GIL functions for genuine multi-threading (#535) (@cjdoris) **Closed issues:** - Libraries loaded via _jll packages are visible inside PythonCall environment. (#519)

## PythonCall v0.9.21 [Diff since v0.9.20](https://github.com/JuliaPy/PythonCall.jl/compare/v0.9.20...v0.9.21) **Merged pull requests:** - making it possible to use another package instead of `pickle` for serialization (e.g. `dill`) (#494) (@schlichtanders) - Timedelta conversion (#499) (@cjdoris) - pyjl (#512) (@cjdoris) - call jl_atexit_hook when atexit (#516) (@dpinol) - check for null in multimedia show functions (#526) (@cjdoris) - Julia formatter (#528) (@cjdoris) **Closed issues:** - single `Jl` type for Julia values (#404) - support "dill" for serialization and deserialization (#424) - Not working with julia version 1.10.1 on linux systems (#464) - datetime.timedelta conversions fail in juliacall (#498) - np.bool_ does not convert to julia Bool (#502) - data-type specific Julia scalars with JuliaCall (#504) - OSError on Apple M2 Chip (#508) - finalizer may never be called (#515) - seg fault from pynew() in vs code REPL (#522)

## PythonCall v0.9.20 [Diff since v0.9.19](https://github.com/JuliaPy/PythonCall.jl/compare/v0.9.19...v0.9.20) **Changes:** * The IPython extension is now automatically loaded upon import if IPython is detected. * JuliaCall now compatible with Julia 1.10.3. * Minimum supported Python version is now 3.8. **Merged pull requests:** - Unittest IPython extension + Automatic loading (#478) (@MilesCranmer) - fix issue 487 (#488) (@cjdoris) **Closed issues:** - numpy functions don't treat Any[...] arrays like Python lists (#486) - `display` broken for parametric structs with `show` referencing the struct parameters (#487) - use existing julia install (#489) - Python: TypeError: Object of type DictValue is not JSON serializable (#491) - Error in calling display(array) (#495)