A worker pool dispatches tasks to a fixed number of threads through a shared channel. Workers pull tasks, process them, and push results to a results channel. This pattern is how http_serve_mt implements its thread pool internally.
// Worker pool — dispatching tasks to N threads via channels // Pool de workers — despachar tareas a N hilos via canales var work_ch: Map = Map.new() var done_ch: Map = Map.new() fn worker() -> int { while true { let task: int = channel_recv(work_ch) if task < 0 { return 0 } // Simulate work: compute square let result: int = task * task channel_send(done_ch, result) } return 0 } fn main() -> int { work_ch = channel_new(20) done_ch = channel_new(20) // Spawn 4 workers let num_workers: int = 4 var handles: Array = [] var i: int = 0 while i < num_workers { let h: int = thread_spawn(worker) handles.push(h) i = i + 1 } // Send 8 tasks let num_tasks: int = 8 i = 0 while i < num_tasks { channel_send(work_ch, i + 1) i = i + 1 } // Collect results i = 0 while i < num_tasks { let result: int = channel_recv(done_ch) print("result: " + int_to_string(result)) i = i + 1 } // Send stop signals i = 0 while i < num_workers { channel_send(work_ch, -1) i = i + 1 } print("all tasks completed") return 0 }
result: 1 result: 4 result: 9 result: 16 result: 25 result: 36 result: 49 result: 64 all tasks completed
How it works
Two channels coordinate the pool: work_ch distributes tasks and done_ch collects results. Both are declared as Map -- the standard channel type in Nyx.
Four worker threads loop on channel_recv(work_ch), computing the square of each received value and sending results to done_ch. A negative task value serves as a poison pill, causing the worker to exit.
The main thread sends 8 tasks (values 1-8), collects all 8 results, then sends 4 stop signals (one per worker). The order of results may vary between runs because workers execute concurrently, but all 8 squares will appear. This is the same pattern used internally by http_serve_mt to handle HTTP requests across a thread pool.