8   - [Tasks](#tasks)
19 …complex part of marl and is responsible for executing tasks and keeping threads running when tasks…
35 …scheduling tasks when there are no [Multi-Threaded-Workers](#multi-threaded-workers) available. Un…
47 ## Tasks  section in `marl::Scheduler`
51 Tasks are scheduled using `marl::schedule()`, and are typically implemented as a lambda:
76 - `work.tasks` - A queue of tasks, yet to be started.
79 - `work.num` - A counter that is kept in sync with `work.tasks.size() + work.fibers.size()`.
83 …e()`, a worker is picked, and the task is placed on to the worker's `work.tasks` queue. The worker…
97 - Calls [`runUntilIdle()`](#marlschedulerworkerrununtilidle) to process all new tasks and fibers. N…
111 1. Resume any unblocked tasks (fibers)
116 …Executing unblocked fibers is prioritized over starting new tasks. This is because new tasks may r…
118 2. Start executing new tasks
120 …fibers have been completed or have become re-blocked, new tasks are taken from the `work.tasks` qu…
122 3. Once there's no more fibers or tasks to execute, `runUntilIdle()` returns.
128 When a worker runs out of tasks to start and fibers to resume, `waitForWork()` is called to block u…
135 - A task becoming enqueued on the `work.tasks` queue.
149 …nd will attempt to steal tasks from randomly picked workers. Because fibers must only be executed …
153 …It is common to have a single task (provider) scheduling many small sub-tasks to the scheduler, wh…
167 Marl allows tasks to block, while keeping threads busy.
172 …s` set and is switched to. This idle fiber when resumed, will continue the role of executing tasks.
173 …3. If none of the above occurs, then a new fiber needs to be created to continue executing tasks. …
191 …ds (`marl::Scheduler::config().workerThreads.count == 0`), then scheduled tasks are queued on to t…
193 …de there are no worker threads, the tasks queued on the STW are not automatically background execu…
215   // calls marl::Scheduler::Worker::run() to run all enqueued tasks. Once the