Concurrency & Async Programming

What it is

Concurrency is arranging for a program to make progress on several things at once. Async programming is one way of doing that without dedicating a thread to each task: work is started, its result is represented by an object you can wait on, and control returns to a loop that runs whatever is ready next. Both exist to keep a machine busy while it waits, and both create the same class of bugs when the sharing rules are unclear.

Why it matters

Concurrency is where correct-looking code fails intermittently and where the failure is hardest to reproduce. A race condition appears under load, on a particular core count, with a particular interleaving — and disappears when you attach a debugger. Async code adds a second failure mode: it looks like ordinary code but the lines around an await do not run in the order written, so state can change underneath you. Learning the small set of rules behind both turns a whole family of unreproducible incidents into something you can reason about on paper.

In this track

1. Threads, locks and races (01-cs-fundamentals/05-concurrency-async-programming/01-threads-locks-and-races) — shared state, mutual exclusion, deadlock and the patterns that avoid both.
2. Async, promises and event loops (01-cs-fundamentals/05-concurrency-async-programming/02-async-promises-and-event-loops) — non-blocking I/O, waiting on a future, and what the loop actually runs.

Prerequisites

None.

Go deeper

• Operating Systems (01-cs-fundamentals/15-operating-systems) — the scheduler and address space that make threads cheap and races possible.
• Debugging (01-cs-fundamentals/11-debugging) — how to reproduce a fault that only happens sometimes.
• Computer Networks (01-cs-fundamentals/03-computer-networks) — waiting for the network is the case that made async programming necessary.