Parallel by Default

This is the chapter where Crow stops resembling Python. In most languages, parallelism is something you add — threads, pools, async, locks. In Crow it's what for already does: iterations run concurrently, on all your cores, and you never wrote a thread.

The design has consequences, and this chapter is honest about them: what the loop gives you, what it asks of you, and the three search builtins that cover the patterns a parallel loop can't.

One boundary to keep in mind as you read: for is about throughput — finishing a computation faster by using every core. Keeping an app responsive while something slow happens is a different job with its own verb, later, which you'll meet in the web chapter (Slow Work and the Live Loop).