IPython — The Interactive Computing Shell for Python

Introduction

IPython replaces the default Python REPL with a far more capable interactive environment. It powers Jupyter kernels and remains the go-to shell for data scientists, researchers, and developers who spend significant time exploring code interactively.

What IPython Does

• Provides an enhanced interactive Python shell with syntax highlighting and auto-indentation
• Supports magic commands (%timeit, %run, %debug) for common workflows
• Offers introspection with ? and ?? to inspect any object or source code
• Integrates with the OS shell via ! prefix for running system commands inline
• Serves as the default kernel backend for Jupyter Notebook and JupyterLab

Architecture Overview

IPython is built around a read-eval-print loop that extends Python's standard code.InteractiveConsole. It uses a two-process model when running as a Jupyter kernel: a frontend (notebook or console) communicates with the IPython kernel over ZeroMQ sockets using the Jupyter messaging protocol. The core shell object manages namespaces, history, and magics as a plugin registry.

Self-Hosting & Configuration

• Install via pip: pip install ipython or conda: conda install ipython
• Configuration lives in ~/.ipython/profile_default/ipython_config.py
• Create custom profiles with ipython profile create <name> for project-specific setups
• Startup scripts placed in ~/.ipython/profile_default/startup/ run automatically on launch
• Extensions load via %load_ext or by adding them to c.InteractiveShellApp.extensions in config

Key Features

• Tab completion powered by Jedi with type-aware suggestions
• Persistent input/output history with SQLite backend and %recall magic
• Built-in debugger integration via %debug post-mortem and %pdb auto-trigger
• Rich display protocol for rendering HTML, images, LaTeX, and SVG inline
• Async/await support at the top level without wrapping in asyncio.run()

Comparison with Similar Tools

• Standard Python REPL — no completion, no history search, no magics; IPython adds all of these
• bpython — good completion and syntax highlighting but lacks magic commands and Jupyter integration
• ptpython — built on prompt_toolkit with a nice UI but smaller ecosystem than IPython
• Jupyter Notebook — uses IPython as its kernel; the notebook adds cell-based documents on top

FAQ

Q: Is IPython the same as Jupyter? A: No. IPython is the Python kernel and interactive shell. Jupyter is the notebook interface that can use IPython (or other language kernels) as a backend.

Q: Can I use IPython in production scripts? A: IPython is designed for interactive use. For production code, use standard Python. You can embed an IPython shell into a running program for debugging with IPython.embed().

Q: Does IPython support other languages? A: IPython itself is Python-only. The Jupyter protocol it helped create supports 100+ language kernels, but those are separate projects.

Q: How do I make IPython my default shell? A: Set your terminal profile or shell alias to launch ipython instead of python. For Jupyter, IPython is already the default Python kernel.

Sources

• https://github.com/ipython/ipython
• https://ipython.readthedocs.io