[][src]Struct futures::task::Task

pub struct Task { /* fields omitted */ }

A handle to a "task", which represents a single lightweight "thread" of execution driving a future to completion.

In general, futures are composed into large units of work, which are then spawned as tasks onto an executor. The executor is responsible for polling the future as notifications arrive, until the future terminates.

This is obtained by the task::current function.


Why does Task not implement Eq and Hash?

A valid use case for Task to implement these two traits has not been encountered.

Usually, this question is asked by someone who wants to store a Task instance in a HashSet. This seems like an obvious way to implement a future aware, multi-handle structure; e.g. a multi-producer channel.

In this case, the idea is that whenever a start_send is called on one of the channel's send handles, if the channel is at capacity, the current task is stored in a set. Then, when capacity is available, a task is removed from the set and notified.

The problem with this strategy is that multiple Sender handles can be used on the same task. In this case, when the second handle is used and the task is stored in a set, there already is an entry. Then, when the first handle is dropped, this entry is cleared, resulting in a dead lock.

See here for more discussion.


impl Task[src]

pub fn notify(&self)[src]

Indicate that the task should attempt to poll its future in a timely fashion.

It's typically guaranteed that, after calling notify, poll will be called at least once subsequently (unless the future has terminated). If the task is currently polling its future when notify is called, it must poll the future again afterwards, ensuring that all relevant events are eventually observed by the future.

pub fn is_current(&self) -> bool[src]


intended to be removed, see docs for details

Returns true when called from within the context of the task.

In other words, the task is currently running on the thread calling the function. Note that this is currently, and has historically, been implemented by tracking an id on every instance of Spawn created. When a Spawn is being polled it stores in thread-local-storage the id of the instance, and then task::current will return a Task that also stores this id.

The intention of this function was to answer questions like "if I notify this task, is it equivalent to task::current().notify()?" The answer "yes" may be able to avoid some extra work to block the current task, such as sending a task along a channel or updating a stored Task somewhere. An answer of "no" typically results in doing the work anyway.

Unfortunately this function has been somewhat buggy in the past and is not intended to be supported in the future. By simply matching id the intended question above isn't accurately taking into account, for example, unpark events (now deprecated, but still a feature). Thus many old users of this API weren't fully accounting for the question it was intended they were asking.

This API continues to be implemented but will in the future, e.g. in the 0.1.x series of this crate, eventually return false unconditionally. It is intended that this function will be removed in the next breaking change of this crate. If you'd like to continue to be able to answer the example question above, it's recommended you use the will_notify_current method.

If you've got questions about this though please let us know! We'd like to learn about other use cases here that we did not consider.


This function will panic if no current future is being polled.

pub fn will_notify_current(&self) -> bool[src]

This function is intended as a performance optimization for structures which store a Task internally.

The purpose of this function is to answer the question "if I notify this task is it equivalent to task::current().notify()". An answer "yes" may mean that you don't actually need to call task::current() and store it, but rather you can simply leave a stored task in place. An answer of "no" typically means that you need to call task::current() and store it somewhere.

As this is purely a performance optimization a valid implementation for this function is to always return false. A best effort is done to return true where possible, but false negatives may happen. Note that this function will not return a false positive, however.


This function will panic if no current future is being polled.

Trait Implementations

impl Debug for Task[src]

impl Clone for Task[src]

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source. Read more

Auto Trait Implementations

impl Send for Task

impl Sync for Task

Blanket Implementations

impl<T> From for T[src]

impl<T, U> TryFrom for T where
    U: Into<T>, 

type Error = Infallible

The type returned in the event of a conversion error.

impl<T, U> TryInto for T where
    U: TryFrom<T>, 

type Error = <U as TryFrom<T>>::Error

The type returned in the event of a conversion error.

impl<T, U> Into for T where
    U: From<T>, 

impl<T> Borrow for T where
    T: ?Sized

impl<T> BorrowMut for T where
    T: ?Sized

impl<T> Any for T where
    T: 'static + ?Sized

impl<T> ToOwned for T where
    T: Clone

type Owned = T