Struct shrev::EventChannel
source · [−]pub struct EventChannel<E> { /* private fields */ }
Expand description
The EventChannel
, which is the central component of shrev
.
How it works
This channel has a ring buffer, which it allocates with an initial capacity. Once allocated, it writes new events into the buffer, wrapping around when it reaches the “end” of the buffer.
However, before an event gets written into the buffer, the channel checks if all readers have read the event which is about to be overwritten. In case the answer is “No”, it will grow the buffer so no events get overwritten.
Readers are stores in the EventChannel
itself, because we need to access
their position in a write, so we can check what’s described above. Thus, you
only get a ReaderId
as a handle.
What do I use it for?
The EventChannel
is basically a single producer, multiple consumer
(“SPMC”) channel. That is, a write
to the channel requires mutable access,
while reading can be done with just an immutable reference. All readers
(consumers) will always get all the events since their last read (or when
they were created, if there was no read yet).
Examples
use std::mem::drop;
use shrev::{EventChannel, ReaderId};
// The buffer will initially be 16 events big
let mut channel = EventChannel::with_capacity(16);
// This is basically with no effect; no reader can possibly observe it
channel.single_write(42i32);
let mut first_reader = channel.register_reader();
// What's interesting here is that we don't check the readers' positions _yet_
// That is because the size of 16 allows us to write 16 events before we need to perform
// such a check.
channel.iter_write(0..4);
// Now, we read 4 events (0, 1, 2, 3)
// Notice how we borrow the ID mutably; this is because logically we modify the reader,
// and we shall not read with the same ID concurrently
let _events = channel.read(&mut first_reader);
// Let's create a second reader; this one will not receive any of the previous events
let mut second_reader = channel.register_reader();
// No event returned
let _events = channel.read(&mut second_reader);
channel.iter_write(4..6);
// Both now get the same two events
let _events = channel.read(&mut first_reader);
let _events = channel.read(&mut second_reader);
// We no longer need our second reader, so we drop it
// This is important, since otherwise the buffer would keep growing if our reader doesn't read
// any events
drop(second_reader);
Implementations
sourceimpl<E> EventChannel<E> where
E: Event,
impl<E> EventChannel<E> where
E: Event,
sourcepub fn with_capacity(size: usize) -> Self
pub fn with_capacity(size: usize) -> Self
Create a new EventChannel
with the given starting capacity.
sourcepub fn would_write(&mut self) -> bool
pub fn would_write(&mut self) -> bool
Returns true
if any reader would observe an additional event.
This can be used to skip calls to iter_write
in case the event
construction is expensive.
sourcepub fn register_reader(&mut self) -> ReaderId<E>
pub fn register_reader(&mut self) -> ReaderId<E>
Register a new reader.
To be able to read events, a reader id is required. This is because
otherwise the channel wouldn’t know where in the ring buffer the
reader has read to earlier. This information is stored in the channel,
associated with the returned ReaderId
.
A newly created ReaderId
will only receive the events written after
its creation.
Once you no longer perform read
s with your ReaderId
, you should
drop it so the channel can safely overwrite events not read by it.
sourcepub fn slice_write(&mut self, events: &[E]) where
E: Clone,
👎 Deprecated: please use iter_write
instead
pub fn slice_write(&mut self, events: &[E]) where
E: Clone,
please use iter_write
instead
Write a slice of events into storage
sourcepub fn iter_write<I>(&mut self, iter: I) where
I: IntoIterator<Item = E>,
I::IntoIter: ExactSizeIterator,
pub fn iter_write<I>(&mut self, iter: I) where
I: IntoIterator<Item = E>,
I::IntoIter: ExactSizeIterator,
Write an iterator of events into storage
sourcepub fn drain_vec_write(&mut self, events: &mut Vec<E>)
pub fn drain_vec_write(&mut self, events: &mut Vec<E>)
Drain a vector of events into storage.
sourcepub fn single_write(&mut self, event: E)
pub fn single_write(&mut self, event: E)
Write a single event into storage.
sourcepub fn read(&self, reader_id: &mut ReaderId<E>) -> EventIterator<'_, E>ⓘNotable traits for StorageIterator<'a, T>impl<'a, T> Iterator for StorageIterator<'a, T> type Item = &'a T;
pub fn read(&self, reader_id: &mut ReaderId<E>) -> EventIterator<'_, E>ⓘNotable traits for StorageIterator<'a, T>impl<'a, T> Iterator for StorageIterator<'a, T> type Item = &'a T;
Read any events that have been written to storage since the last read
with reader_id
(or the creation of the ReaderId
, if it hasn’t read
yet).
Note that this will advance the position of the reader regardless of
what you do with the iterator. In other words, calling read
without iterating the result won’t preserve the events returned. You
need to iterate all the events as soon as you got them from this
method. This behavior is equivalent to e.g. Vec::drain
.
Trait Implementations
sourceimpl<E: Debug> Debug for EventChannel<E>
impl<E: Debug> Debug for EventChannel<E>
Auto Trait Implementations
impl<E> !RefUnwindSafe for EventChannel<E>
impl<E> Send for EventChannel<E> where
E: Send,
impl<E> Sync for EventChannel<E> where
E: Sync,
impl<E> Unpin for EventChannel<E> where
E: Unpin,
impl<E> !UnwindSafe for EventChannel<E>
Blanket Implementations
sourceimpl<T> BorrowMut<T> for T where
T: ?Sized,
impl<T> BorrowMut<T> for T where
T: ?Sized,
const: unstable · sourcefn borrow_mut(&mut self) -> &mut T
fn borrow_mut(&mut self) -> &mut T
Mutably borrows from an owned value. Read more