Self-draining for-loop job used by ThreadPool::ParallelFor. More...

#include <ThreadPool.hpp>

Inheritance diagram for Foundation::Core::ParallelForJob< Fn >:

Public Member Functions
void	Execute (size_t workerId) noexcept override

Public Member Functions inherited from Foundation::Core::Job
virtual	~Job ()=default

Public Attributes
Fn *	fn {nullptr}

size_t	total {0}

Atomic< size_t >	cursor {0}

Atomic< size_t >	remaining {0}

Detailed Description

template<typename Fn>
struct Foundation::Core::ParallelForJob< Fn >

Self-draining for-loop job used by ThreadPool::ParallelFor.

One instance is co-invoked across N workers (+ the calling thread) via ThreadPool::CoInvoke; each pulls indices from a shared atomic cursor until the range is exhausted, invoking fn per index. It lives on the caller's stack — no heap allocation, no future. fn is invoked concurrently as fn(i) / fn(i, workerId), so it must be thread-safe: shared captures read-only, writes disjoint per index, any scratch keyed by workerId (a worker never runs two invocations at once, so per-worker scratch is race-free). Per-index granularity is intentional — a job that wants coarser work batches itself by choosing count.