core.async Channels and Go Blocks

Atoms and refs manage shared state. core.async manages communication between independent processes using channels, inspired by CSP.

Pull in the core functions you will use.

A channel is a queue that processes use to hand off values. Create one with chan.

A go block runs its body asynchronously on a thread pool, parking instead of blocking at channel operations.

Inside a go block, >! puts a value and <! takes one. Both park until the other side is ready.

Outside go blocks, use the double-bang forms >!! and <!!, which block the real thread. Useful at the REPL or in main.

Give a channel a buffer so puts do not park until the buffer is full. This decouples fast producers from slower consumers.

alts! waits on several channels at once and proceeds with whichever is ready first, enabling timeouts and racing.

Closing a channel signals no more values. Takes then return nil, which consumers use as a stop signal.

Wire go blocks together with channels to form a pipeline: each stage takes from one channel, transforms, and puts onto the next. State stays local to each process.

Use atoms/refs for shared mutable state; use channels to coordinate the flow of work between processes. They solve different problems and often combine.

Test your core.async knowledge.

You learned channels, go blocks, parking puts/takes, blocking variants, buffers, alts! for selection and timeouts, closing channels, and how channels complement atoms and refs.