This branch is used Google-internally to build thousands of extensions that are deployed to production and is also regularly tested with all CLANG sanitizers. However, feature stability guarantees are currently limited in this way:
- pywrapcc is meant to be a fully backward-compatible superset of pybind11 (master). If not, that is not intended, please let us know by creating an issue.
- The main driving force for adding new features is the PyCLIF-pybind11 integration work (for Googlers: go/pyclif_pybind11_fusion). Until this work is completed, it is impractical for us to take external use cases into account when evolving new features. This is because we can globally test changes fairly easily internally, but not externally.
- After the PyCLIF-pybind11 work is completed we will commit to greater feature stability and update the documentation.
Outstanding new features: py::native_enum (google#30005), py::return_value_policy_pack (google#30011), enhanced pybind11/functional.h API (google#30022)
This pywrapcc repo originated from the pybind11 smart_holder branch. It was created with two important goals in mind:
- Sustained innovation and proactive bug fixes.
- Sustained continuity.
With original pybind11, these two goals are in a conflict. In the early days of pybind11 this was not so much of a problem, but this has been changing with growing adoption.
Regarding goal 1: pybind11 has two serious long-standing bugs (pybind/pybind11#1138, pybind/pybind11#1333) that have never been fixed on the master branch, but were fixed on the smart_holder branch in early-mid 2021.
Regarding goal 2: The original and still current pybind11::smart_holder implementation is a compromise solution that avoids an "ABI break", concretely, changes to the pybind11::details::internals struct. Each time the internals are changed, the PYBIND11_INTERNALS_VERSION needs to be incremented, cutting off interoperability with existing PyPI wheels based on pybind11, without giving any hint about this problem at runtime. In the meantime, two other PRs were merged on the pybind11 master branch that require changes to the internals, PRs pybind/pybind11#3275 and pybind/pybind11#4254. To avoid ABI breaks, these PRs were effectively held back behind #ifdef s. This problem came to a breaking point with PR pybind/pybind11#4329, for which hiding the new feature behind #ifdef s is not a practical option.
Obviously, neither repeatedly breaking interoperability with existing PyPI wheels, nor holding back bug fixes and new features, is desirable. Therefore PR pybind/pybind11#4329 was extended to include a generalization of the internals approach, under the name cross_extension_shared_state. The fundamental difference is that established shared state is left untouched, and new shared states are added as needed, largely resolving the conflict between innovation & continuity. The price to pay is added complexity managing the evolving shared states, but that is assumed to be a relatively small extra effort for a few developers, resulting in a big usability gain for a much larger number of users. Ultimately, this is just the familiar innovate-deprecate-cleanup life cycle typical for many (all?) long-term successful major projects (e.g. C++, Python). Even pywrapcc developers are likely to see this as a win worth paying a price for, because they are more free to innovate.
A direct consequence of goal 2. is that the C++ pybind11 namespace cannot abruptly be changed, because renaming would break both API and ABI compatibility. The intent is to change the API gradually, driven primarily by code health and innovation-related refactoring needs, more than arbitrary "let's change some names" decisions. ABI compatibility will be phased out gradually, on time scales similar to Python EOL policies. Eventually there may be little or no "pybind11" left in pywrapcc, but this will certainly take some time.
pybind11 is a lightweight header-only library that exposes C++ types in Python and vice versa, mainly to create Python bindings of existing C++ code. Its goals and syntax are similar to the excellent Boost.Python library by David Abrahams: to minimize boilerplate code in traditional extension modules by inferring type information using compile-time introspection.
The main issue with Boost.Python—and the reason for creating such a similar project—is Boost. Boost is an enormously large and complex suite of utility libraries that works with almost every C++ compiler in existence. This compatibility has its cost: arcane template tricks and workarounds are necessary to support the oldest and buggiest of compiler specimens. Now that C++11-compatible compilers are widely available, this heavy machinery has become an excessively large and unnecessary dependency.
Think of this library as a tiny self-contained version of Boost.Python with everything stripped away that isn't relevant for binding generation. Without comments, the core header files only require ~4K lines of code and depend on Python (3.6+, or PyPy) and the C++ standard library. This compact implementation was possible thanks to some of the new C++11 language features (specifically: tuples, lambda functions and variadic templates). Since its creation, this library has grown beyond Boost.Python in many ways, leading to dramatically simpler binding code in many common situations.
Tutorial and reference documentation is provided at pybind11.readthedocs.io. A PDF version of the manual is available here. And the source code is always available at github.com/pybind/pybind11.
pybind11 can map the following core C++ features to Python:
- Functions accepting and returning custom data structures per value, reference, or pointer
- Instance methods and static methods
- Overloaded functions
- Instance attributes and static attributes
- Arbitrary exception types
- Enumerations
- Callbacks
- Iterators and ranges
- Custom operators
- Single and multiple inheritance
- STL data structures
- Smart pointers with reference counting like
std::shared_ptr - Internal references with correct reference counting
- C++ classes with virtual (and pure virtual) methods can be extended in Python
In addition to the core functionality, pybind11 provides some extra goodies:
- Python 3.6+, and PyPy3 7.3 are supported with an implementation-agnostic interface (pybind11 2.9 was the last version to support Python 2 and 3.5).
- It is possible to bind C++11 lambda functions with captured variables. The lambda capture data is stored inside the resulting Python function object.
- pybind11 uses C++11 move constructors and move assignment operators whenever possible to efficiently transfer custom data types.
- It's easy to expose the internal storage of custom data types through Pythons' buffer protocols. This is handy e.g. for fast conversion between C++ matrix classes like Eigen and NumPy without expensive copy operations.
- pybind11 can automatically vectorize functions so that they are transparently applied to all entries of one or more NumPy array arguments.
- Python's slice-based access and assignment operations can be supported with just a few lines of code.
- Everything is contained in just a few header files; there is no need to link against any additional libraries.
- Binaries are generally smaller by a factor of at least 2 compared to equivalent bindings generated by Boost.Python. A recent pybind11 conversion of PyRosetta, an enormous Boost.Python binding project, reported a binary size reduction of 5.4x and compile time reduction by 5.8x.
- Function signatures are precomputed at compile time (using
constexpr), leading to smaller binaries. - With little extra effort, C++ types can be pickled and unpickled similar to regular Python objects.
- Clang/LLVM 3.3 or newer (for Apple Xcode's clang, this is 5.0.0 or newer)
- GCC 4.8 or newer
- Microsoft Visual Studio 2017 or newer
- Intel classic C++ compiler 18 or newer (ICC 20.2 tested in CI)
- Cygwin/GCC (previously tested on 2.5.1)
- NVCC (CUDA 11.0 tested in CI)
- NVIDIA PGI (20.9 tested in CI)
This project was created by Wenzel Jakob. Significant features and/or improvements to the code were contributed by Jonas Adler, Lori A. Burns, Sylvain Corlay, Eric Cousineau, Aaron Gokaslan, Ralf Grosse-Kunstleve, Trent Houliston, Axel Huebl, @hulucc, Yannick Jadoul, Sergey Lyskov, Johan Mabille, Tomasz Miąsko, Dean Moldovan, Ben Pritchard, Jason Rhinelander, Boris Schäling, Pim Schellart, Henry Schreiner, Ivan Smirnov, Boris Staletic, and Patrick Stewart.
We thank Google for a generous financial contribution to the continuous integration infrastructure used by this project.
See the contributing guide for information on building and contributing to pybind11.
pybind11 is provided under a BSD-style license that can be found in the LICENSE file. By using, distributing, or contributing to this project, you agree to the terms and conditions of this license.