Programming Fundamentals › Memory & Runtime
Compiled vs Interpreted
Translating to machine code ahead of time vs executing source at runtime.
Also known as: compiled language, interpreted language, JIT, bytecode
Programs written in a high-level language have to end up as instructions the CPU can run. There are two broad ways to get there.
- Compiled: a compiler translates the whole program into machine code ahead of time. You run the resulting executable. C, C++, Go, Rust.
- Interpreted: an interpreter reads and executes the source (or an intermediate form) while the program runs. Older Python, Ruby, shell scripts.
Compiled: source ──compile──► machine code ──► run (fast, no source needed at runtime)
Interpreted: source ──────────► interpreter reads and runs it, line by lineTrade-offs
| Compiled ahead of time | Interpreted | |
|---|---|---|
| Speed at runtime | usually fastest | usually slower |
| Startup | fast | slower (needs the interpreter) |
| Development loop | a compile step | run directly, quick experiments |
| Errors | many caught at compile time | found when that line runs |
| Portability | one binary per platform | the same source runs wherever the interpreter does |
It’s blurrier than the labels suggest
- Bytecode: Java and C# compile to bytecode, which a virtual machine runs. Python compiles source to bytecode too, then interprets that.
- JIT (just-in-time) compilation: JavaScript engines, the JVM and others watch which code runs often and compile it to machine code on the fly.
- Transpiling: TypeScript and others compile to JavaScript.
So “Python is interpreted” and “Java is compiled” are simplifications. Whether a language is compiled or interpreted is a property of its implementation, not the language itself.
Why you’d care
- Deployment: a compiled binary can be shipped alone, while an interpreted app needs the runtime and its dependencies installed.
- Performance tuning: knowing a language is JIT-compiled explains warm-up times and why benchmarks need repeated runs.
- Error timing: compile errors stop you early, while runtime errors may reach users.
- It’s separate from static vs dynamic typing (static vs dynamic).