The language
A language you can read on the first draft.
Python is a language for programs a person can follow, from a short script to a service that calls a large library. The lessons teach the syntax. This page is the reason the syntax looks the way it does.
Origins
Guido van Rossum began Python in December 1989, at CWI in the Netherlands, over a Christmas break. He wanted a successor to ABC, a teaching language he had worked on, that would also feel at home to people writing C and Unix tools. The name comes from Monty Python's Flying Circus. The snake came later, as a mascot.
The first public release, 0.9.0, went out in February 1991. Python 1.0 followed in January 1994. Python 2.0 arrived in October 2000. Python 3.0, in December 2008, was a deliberate break: print became a function, text and bytes split apart, and some old habits were removed so the language could stay coherent.
Python 2.7 was maintained until 1 January 2020. The programs in Serpens use syntax from Python 3.10 onward, including match and int | None. Try it opens Python Tutor, which runs them as Python 3.11.
The Python Software Foundation, created in 2001, holds the intellectual property and runs PyPI, the index of packages. Guido was the language's benevolent dictator for life until July 2018. A steering council governs it now. Indentation is part of the syntax, which is why a block in these lessons is a change of indent and not a pair of braces.
Beside other languages
Python shares habits with languages you may already know, and it spends its strictness on indentation and on errors at the line that runs.
| Python | Rust | JavaScript | Java | |
|---|---|---|---|---|
| Types | Decided as the program runs. Annotations are hints. | Checked before the program runs | Dynamic. TypeScript checks, then erases the types | Checked before the program runs |
| Memory | Garbage collected. Names refer to objects. | Freed when the owner ends. No garbage collector | Garbage collected | Garbage collected, on a virtual machine |
| Absence and failure | None and exceptions | Option and Result | undefined and exceptions | null and exceptions |
| Runs as | An interpreter, or a bytecode virtual machine | Native code, or WebAssembly | A browser, Node, or another JS engine | The Java virtual machine |
Rust
Both reward a program you can read. Rust gets there by refusing to build until the types, the owners, and the borrows agree. Python gets there with names, indentation, and a large library. A wrong type in Python usually appears when that line runs. Rust is the tighter program when the work is the CPU, the memory, or a bug you cannot afford later. Python is the faster first draft when the program spends its time waiting, or calling a library that already did the fast part.
JavaScript
Both are dynamic, both are garbage collected, and both have a large package culture, PyPI and npm. JavaScript is the language of the page. It runs in the browser and in Node. TypeScript adds types and then removes them: what runs is still JavaScript. Python runs as an interpreter or a bytecode virtual machine. It is the usual language of notebooks, automation, and the scripts around a web system. The page itself stays JavaScript.
Java
Both are used for programs that live for years, and both can say what a type is for. Java checks those types before the program runs, and it runs on a virtual machine. Almost any reference can be null, and failure is often an exception. Python checks a value when the line runs. Annotations can be checked by a separate tool, and the interpreter does not require that check. A team with a large Java codebase will often ship sooner by staying there. Python pays that choice back when the work is a script, a service with a small team, or a library the scientist or the operator already knows.
Why someone reaches for it
People choose Python when the next reader matters as much as the next release.
- The first version should be readable, and a teammate should be able to change it next week.
- The work is a script, a notebook, a service, or a call into a library, and PyPI already has that library.
- A wrong type can wait until that line runs, because this program is not the piece that must be rejected at build time.
- The slow part can live in a library written in C or Rust, and Python can be the language that calls it.
A program that must control every byte, avoid garbage-collection pauses, or reject a mistake before it starts is usually happier in Rust or Java. Python is the language for the part of the system a person has to read, glue, and change.
Where it has a following
The following is in teaching, in science, in automation, and in the services that sit in front of faster libraries.
- Teaching. Python is often the first language in a classroom, because a small program is a small file and the error names the line.
- Science and data. NumPy, pandas, and Jupyter are the usual tools. A notebook is a Python program with the results kept beside the code.
- The web. Django, Flask, and FastAPI are Python. Instagram grew up on Django.
- Automation. Operators and builders use Python for the glue around files, APIs, and deployments. Dropbox was written largely in Python.
- Machine learning. PyTorch and scikit-learn present themselves as Python. Much of the heavy arithmetic is in C, C++, or a GPU kernel. The training script, the notebook, and the experiment are Python.
YouTube's early stack was Python as well. The pattern is the same in each of these: Python is the language people write, and faster code sits underneath when the work needs it.
Start at the top when you want the syntax those programs are written in.