1 min read

pyssg.core.scheduler

Phase scheduler.

Runs a phase's work over many nodes. Because every processing unit is pure in its declared inputs, the order of completion cannot affect results, which is what makes parallelization sound. This M5 implementation is a deterministic sequential map; it preserves input order so output is reproducible, and the interface leaves room to swap in a thread/process pool later without touching callers.

class Scheduler

Deterministic map over node ids within a single phase.

Scheduler.map(self, ids: list[NodeId], fn: Callable[[NodeId], T]) -> list[T]

Apply fn to each id, returning results in input order.