document JULIA_PYTHONCALL_PICKLE

Author: Christopher Doris

docs/src/compat.md

@@ -9,6 +9,12 @@ functions to bridge a gap. We aim to keep these as minimal as possible.
 
 Whenever a Python exception is displayed by Julia, `sys.last_traceback` and friends are set. This allows the post-mortem debugger `pdb.pm()` to work. Disable by setting `PythonCall.CONFIG.auto_sys_last_traceback = false`.
 
+## Julia standard library
+
+Python objects can be serialised with the [`Serialization`](https://docs.julialang.org/en/v1/stdlib/Serialization/) stdlib.
+This uses [`pickle`](https://docs.python.org/3/library/pickle.html) library under the hood.
+You can opt into using [`dill`](https://pypi.org/project/dill/) instead by setting the environment variable `JULIA_PYTHONCALL_PICKLE="dill"`.
+
 ## Tabular data / Pandas
 
 The abstract type [`PyTable`](@ref) is for wrapper types around Python tables, providing the

docs/src/releasenotes.md

@@ -1,5 +1,8 @@
 # Release Notes
 
+## Unreleased
+* `Serialization.serialize` can use `dill` instead of `pickle` by setting the env var `JULIA_PYTHONCALL_PICKLE=dill`.
+
 ## 0.9.20 (2024-05-01)
 * The IPython extension is now automatically loaded upon import if IPython is detected.
 * JuliaCall now compatible with Julia 1.10.3.

src/Compat/serialization.jl

@@ -2,11 +2,13 @@
 #
 # We use pickle to serialise Python objects to bytes.
 
+_pickle_module() = pyimport(get(ENV, "JULIA_PYTHONCALL_PICKLE", "pickle"))
+
 function serialize_py(s, x::Py)
     if pyisnull(x)
         serialize(s, nothing)
     else
-        b = pyimport(get(ENV, "JULIA_PYTHONCALL_PICKLE", "pickle")).dumps(x)
+        b = _pickle_module().dumps(x)
         serialize(s, pybytes_asvector(b))
     end
 end
@@ -16,7 +18,7 @@ function deserialize_py(s)
     if v === nothing
         pynew()
     else
-        pyimport(get(ENV, "JULIA_PYTHONCALL_PICKLE", "pickle")).loads(pybytes(v))
+        _pickle_module().loads(pybytes(v))
     end
 end