enqueueLocal

Add a task to the work queue pinned to this peer — the headline pinned-execution affordance.

Pinning decouples who-runs from the consistent-hash ring: this node, and only this node, claims and runs the task. No other peer ever executes it, and if this peer is partitioned the task stays pending (it does not re-home to a survivor) until this peer returns. This makes inherently data-local workloads expressible — e.g. federated learning, where only the data owner can run its own training step on its private local data — and doubles as task affinity / sticky placement.

The task is enqueued with TaskDescriptor.pinnedOwner forced to selfId; any owner the caller may have set on descriptor is overridden so "local" always means this peer. Nothing else about the envelope changes — the fair-share TaskDescriptor.lane, the location TaskDescriptor.affinity, the trace context and the args all ride through the pin untouched, so a pinned task is admission-gated and traced exactly like a ring-assigned one. It then travels and is claimed exactly like any free-path task via enqueue — the pin is just the descriptor field the ring-owner filter consults (see effectiveOwner).

Parameters