04·dis and CPython bytecode

Introduction

Python is an INTERPRETED language, but it does not execute source code line by line. It first compiles it into BYTECODE — a sequence of instructions for the CPython virtual machine. Only that bytecode is executed by the evaluation loop (_PyEval_EvalFrameDefault in ceval.c).

The stdlib dis module lets you inspect that bytecode. Type dis.dis(func) and you see instructions: LOAD_FAST, BINARY_ADD, CALL_FUNCTION, RETURN_VALUE. It is a window into what Python actually does — it often explains micro-optimisations (why locals beat globals, why list-comps beat for-loops, why += on strings can be slow).

This lesson covers: reading dis.dis output, the most important opcodes (LOAD_FAST/LOAD_GLOBAL, BUILD_LIST, JUMP_*, CALL), compilation to .pyc in __pycache__, peephole optimisations, the 3.10 → 3.11 differences (PEP 659 specialisation, adaptive interpreter, "inline caches") and using dis to validate optimisations.