code

Code

Source code samples: snippets to paste into a demo, files to feed a syntax highlighter, a linter, a formatter, or a parser, and small programs to show a language doing something.

Layout

One subfolder per language, named after the language in lowercase:

Folder	Language	Extensions
c/	C	.c, .h
cpp/	C++	.cpp, .hpp
csharp/	C#	.cs
dart/	Dart	.dart
go/	Go	.go
java/	Java	.java
javascript/	JavaScript	.js, .mjs, .cjs
kotlin/	Kotlin	.kt
php/	PHP	.php
python/	Python	.py
ruby/	Ruby	.rb
rust/	Rust	.rs
shell/	Shell	.sh
swift/	Swift	.swift
typescript/	TypeScript	.ts

A language that has no folder yet gets one, named the same way. The folder is the language, never the framework — a React component is a sample of JavaScript or TypeScript, and says react in its file name instead.

Naming

Code is the one place in this repository where names follow the language's own convention rather than the repository's kebab-case rule, because some languages enforce theirs: HelloWorld.java has to match its class, and hello_world.py is what a Python module looks like.

Everything else still holds:

• Name the file after what the sample demonstrates: fizzbuzz.py, BinarySearch.java, read-file-async.js, struct_traits.rs.
• Add the qualifier to the name when the point is a variant: sort-quick.go, sort-merge.go, fizzbuzz-recursive.py.
• Mark a sample that is meant to be broken with -invalid, -malformed, or -broken, the same way datasets do. A file that will not compile on purpose has to say so.

Two sizes

Each language folder holds samples of two sizes, and both are wanted:

• Snippets — one construct each, a screen or two long: closures-counter.js, SwitchExpressions.java, pattern_matching.rb. These are what a syntax highlighter, a formatter, or a tutorial wants.
• Programs — a few hundred to a thousand lines, doing something whole: lisp_interpreter.py, MazeSolver.java, query-builder.php. These are what an indexer, a static analyser, a diff viewer, or a context-window test wants, and they are the only samples that show a language's constructs used together.

Two programs per language is the target. They are named after what they do, not after their size — a reader can tell them apart by opening one.

What belongs here

• Self-contained files. A sample should be readable on its own and run, compile, or at least parse without a project around it. No build scripts, no lockfiles, no vendored dependencies. This holds for the long programs too: each is one file with an entry point and, where the language has a convention for it, its own tests.
• Plain standard-library code, unless the sample exists to show a specific library.
• A header comment on a program saying what it does and how to run it. A snippet rarely needs one; a program always does.

Markup and stylesheets are documents rather than programs, so .html and .css samples live in documents/, and SQL lives in datasets/sql/ with the rest of the data.

Sources

Code written for this repository is covered by its LICENSE. A sample that comes from somewhere else keeps its own license, and is listed here with its source, author, and terms before it is added. Nothing has been added yet.