Lesson 03 · Spawn and task ids - Build a Green-Thread Scheduler

Spawn and task ids

To talk about "task 1 ran, then task 2" you need tasks to have stable identities. Today the scheduler gains Spawn - it wraps a step function into a task, stamps it with the next id, and drops it on the run queue.

The goal

Give the scheduler a Spawn that assigns increasing ids and enqueues the new task.

Start here - the target

TO DO

Scenario: Spawning stamps ids and enqueues

Givena fresh scheduler

Whentwo step functions are spawned

Thenthe first task gets id 1 and the second gets id 2

Andthe run queue length is 2 after both spawns

Background

A green thread needs a name so the scheduler - and every test - can refer to it. We use a simple monotonic id: the first task spawned is 1, the next is 2, and so on. Ids never repeat, so a trace like [1, 2, 1] reads unambiguously as “task 1 ran, task 2 ran, task 1 ran again.”

Spawn is the public door into the scheduler: hand it a step function and it becomes a live task at the back of the run queue, ready to run when its turn comes. Add the id field to Task now. That is all spawning does today - it does not run anything. Running is the scheduler loop, and that is next.

Make it work

type Scheduler struct {
  rq     runq
  nextID int
}
func (s *Scheduler) Spawn(step func() Status) *Task {
  s.nextID++                       // ids start at 1
  t := &Task{id: s.nextID, step: step}
  s.rq.enqueue(t)
  return t
}
// add an id field to Task

CheckpointDONE

The scheduler can spawn tasks with stable, increasing ids. Commit and stop here.