Documentation

A monad that can perform concurrent or parallel IO operations. Streams that can be composed concurrently require the underlying monad to be MonadAsync.

Since: 0.1.0 (Streamly)

Since: 0.8.0

Identify the type of the SVar. Two computations using the same style can be scheduled on the same SVar.

Adapt the stream state from one type to another.

When we run computations concurrently, we completely isolate the state of the concurrent computations from the parent computation. The invariant is that we should never be running two concurrent computations in the same thread without using the runInIO function. Also, we should never be running a concurrent computation in the parent thread, otherwise it may affect the state of the parent which is against the defined semantics of concurrent execution.

An SVar or a Stream Var is a conduit to the output from multiple streams running concurrently and asynchronously. An SVar can be thought of as an asynchronous IO handle. We can write any number of streams to an SVar in a non-blocking manner and then read them back at any time at any pace. The SVar would run the streams asynchronously and accumulate results. An SVar may not really execute the stream completely and accumulate all the results. However, it ensures that the reader can read the results at whatever paces it wants to read. The SVar monitors and adapts to the consumer's pace.

An SVar is a mini scheduler, it has an associated workLoop that holds the stream tasks to be picked and run by a pool of worker threads. It has an associated output queue where the output stream elements are placed by the worker threads. A outputDoorBell is used by the worker threads to intimate the consumer thread about availability of new results in the output queue. More workers are added to the SVar by fromStreamVar on demand if the output produced is not keeping pace with the consumer. On bounded SVars, workers block on the output queue to provide throttling of the producer when the consumer is not pulling fast enough. The number of workers may even get reduced depending on the consuming pace.

New work is enqueued either at the time of creation of the SVar or as a result of executing the parallel combinators i.e. <| and <|> when the already enqueued computations get evaluated. See joinStreamVarAsync.

Events that a child thread may send to a parent thread.

Sorting out-of-turn outputs in a heap for Ahead style streams

This function is used by the producer threads to queue output for the consumer thread to consume. Returns whether the queue has more space.

In contrast to pushWorker which always happens only from the consumer thread, a pushWorkerPar can happen concurrently from multiple threads on the producer side. So we need to use a thread safe modification of workerThreads. Alternatively, we can use a CreateThread event to avoid using a CAS based modification.

Specifies the stream yield rate in yields per second (Hertz). We keep accumulating yield credits at rateGoal. At any point of time we allow only as many yields as we have accumulated as per rateGoal since the start of time. If the consumer or the producer is slower or faster, the actual rate may fall behind or exceed rateGoal. We try to recover the gap between the two by increasing or decreasing the pull rate from the producer. However, if the gap becomes more than rateBuffer we try to recover only as much as rateBuffer.

rateLow puts a bound on how low the instantaneous rate can go when recovering the rate gap. In other words, it determines the maximum yield latency. Similarly, rateHigh puts a bound on how high the instantaneous rate can go when recovering the rate gap. In other words, it determines the minimum yield latency. We reduce the latency by increasing concurrency, therefore we can say that it puts an upper bound on concurrency.

If the rateGoal is 0 or negative the stream never yields a value. If the rateBuffer is 0 or negative we do not attempt to recover.

Since: 0.5.0 (Streamly)

Since: 0.8.0

Fork a thread to run the given computation, installing the provided exception handler. Lifted to any monad with 'MonadBaseControl IO m' capability.

TODO: the RunInIO argument can be removed, we can directly pass the action as "mrun action" instead.

fork lifted to any monad with 'MonadBaseControl IO m' capability.

Fork a thread that is automatically killed as soon as the reference to the returned threadId is garbage collected.

Write a stream to an SVar in a non-blocking manner. The stream can then be read back from the SVar using fromSVar.