Interfacing Python with Other Languages

One Language, Several Runtimes

When people say that they are using Python, they may be referring to two related but different things. Python is a language specification: a set of rules for writing Python code. A Python implementation is software that reads and executes code written according to those rules. This distinction explains how several different programs can implement the same language while targeting different platforms and use cases.

Following Code to the Machine

A useful way to trace Python is to start with Python code, place the language specification beside it as the set of rules being followed, and then identify the implementation that executes the code. The implementation is the platform-facing part of this path. Different implementations can connect the same language specification to different operating systems, runtimes, or optimization strategies.

CPython as the Reference Point

CPython is the reference implementation and the most widely used Python interpreter. When you install Python from python.org and run python in a terminal, you are using CPython. It is written in the C programming language, which explains the C in its name. The source material also describes CPython as the Classical Python interpreter because it is the original reference implementation: the implementation against which other implementations are measured and that first receives new language features and changes.

CPython is described as fast and efficient because it is written in C, a compiled, statically typed language. It is also portable across Windows, macOS, Linux, and many other operating systems. Its reference status, broad portability, performance, and large ecosystem make it the normal starting choice for most Python projects.

Choosing an Implementation

Multiple implementations exist because one implementation cannot optimize every goal equally. CPython is the general default and reference point. Jython is designed for Java integration. IronPython is designed for .NET integration. PyPy focuses on an alternative implementation strategy and performance optimization through just-in-time compilation.

Bridging Java and .NET

Jython and IronPython are platform-specific implementations. Jython runs Python on the Java platform, within the Java Virtual Machine. Python code can instantiate Java classes, call their methods, and pass data back and forth. IronPython provides the corresponding connection for the .NET runtime, allowing Python code to interact with .NET libraries and classes, including libraries written in C#, the primary language of .NET.

PyPy and Self-Implementation

PyPy is an implementation of Python written in Python itself, more precisely in a restricted subset called RPython. RPython is translated into C and compiled. This design is not a claim that Python runs without any implementation. Instead, it expresses the interpreter's parsing and execution logic in a Python-based language before translating that logic into a compiled form.

The main development benefit is that PyPy's interpreter logic can be changed using a dynamic language rather than a language such as C or Java. The source describes this as making experimentation faster and the codebase more accessible to the broader Python community. PyPy also includes a just-in-time compiler. For certain workloads, especially long-running programs with tight loops, this can make Python code significantly faster than it is in CPython.

Mistakes in Implementation Choices

• Treating Python and CPython as exactly the same thing

  Python is the specification, while CPython is one implementation of that specification.

  Fix: Use Python to refer to the language and CPython to refer to the reference implementation written in C.

• Choosing Jython or IronPython only because they are alternative Python names

  These implementations exist to connect Python with specific host ecosystems.

  Fix: Choose Jython for deep Java integration and IronPython for integration with the .NET environment.

• Assuming that PyPy is always faster

  The source describes PyPy's performance advantages for certain workloads, especially long-running programs with tight loops.

  Fix: Benchmark performance-critical code with PyPy before deciding.

• Relying on implementation-specific features when portability matters

  Such dependencies reduce the ability to move the code between implementations.

  Fix: Write standard Python code that does not rely on implementation-specific features.

Making the Choice

Start with CPython in most cases because it is the reference implementation, the most widely used implementation, and the one with the largest ecosystem of third-party libraries. Move to a different implementation when the project has a clear requirement: use Jython for deep Java integration, IronPython for .NET integration, or PyPy when measured performance benefits justify the change. Keeping code standard and portable preserves the ability to reconsider the implementation later.

Key Takeaways

1. Python is a language specification; a Python implementation is the software that executes Python code.
2. CPython is the C-written reference implementation and the most widely used Python interpreter.
3. Jython and IronPython connect Python with Java and .NET platforms, respectively.
4. PyPy is written in RPython, translated into C, and includes a JIT compiler that may improve performance for certain workloads.
5. Use CPython by default, choose a platform-specific implementation for integration needs, and benchmark PyPy when performance is the reason for switching.