pub trait Service<Req> {
type Response;
type Error;
// Required method
fn call(
&self,
req: Req,
ctx: ServiceCtx<'_, Self>
) -> impl Future<Output = Result<Self::Response, Self::Error>>;
// Provided methods
fn poll_ready(&self, cx: &mut Context<'_>) -> Poll<Result<(), Self::Error>> { ... }
fn poll_shutdown(&self, cx: &mut Context<'_>) -> Poll<()> { ... }
fn map<F, Res>(self, f: F) -> ServiceChain<Map<Self, F, Req, Res>, Req>
where Self: Sized,
F: Fn(Self::Response) -> Res { ... }
fn map_err<F, E>(self, f: F) -> ServiceChain<MapErr<Self, F, E>, Req>
where Self: Sized,
F: Fn(Self::Error) -> E { ... }
}
Expand description
An asynchronous function of Request
to a Response
.
The Service
trait represents a request/response interaction, receiving requests and returning
replies. You can think about service as a function with one argument that returns some result
asynchronously. Conceptually, the operation looks like this:
async fn(Request) -> Result<Response, Error>
The Service
trait just generalizes this form. Requests are defined as a generic type parameter
and responses and other details are defined as associated types on the trait impl. Notice that
this design means that services can receive many request types and converge them to a single
response type.
Services can also have mutable state that influence computation by using a Cell
, RefCell
or Mutex
. Services intentionally do not take &mut self
to reduce overhead in the
common cases.
Service
provides a symmetric and uniform API; the same abstractions can be used to represent
both clients and servers. Services describe only transformation operations which encourage
simple API surfaces. This leads to simpler design of each service, improves test-ability and
makes composition easier.
struct MyService;
impl Service<u8> for MyService {
type Response = u64;
type Error = Infallible;
async fn call(&self, req: u8, ctx: ServiceCtx<'_, Self>) -> Result<Self::Response, Self::Error> {
Ok(req as u64)
}
}
Sometimes it is not necessary to implement the Service trait. For example, the above service
could be rewritten as a simple function and passed to fn_service
.
async fn my_service(req: u8) -> Result<u64, Infallible>;
Service cannot be called directly, it must be wrapped to an instance of [Pipeline``] or by using
ctx` argument of the call method in case of chanined services.
Required Associated Types§
Required Methods§
sourcefn call(
&self,
req: Req,
ctx: ServiceCtx<'_, Self>
) -> impl Future<Output = Result<Self::Response, Self::Error>>
fn call( &self, req: Req, ctx: ServiceCtx<'_, Self> ) -> impl Future<Output = Result<Self::Response, Self::Error>>
Process the request and return the response asynchronously.
This function is expected to be callable off-task. As such, implementations of call
should take care to not call poll_ready
. Caller of the service verifies readiness,
Only way to make a call
is to use ctx
argument, it enforces readiness before calling
service.
Provided Methods§
sourcefn poll_ready(&self, cx: &mut Context<'_>) -> Poll<Result<(), Self::Error>>
fn poll_ready(&self, cx: &mut Context<'_>) -> Poll<Result<(), Self::Error>>
Returns Ready
when the service is able to process requests.
If the service is at capacity, then Pending
is returned and the task is notified when
the service becomes ready again. This function is expected to be called while on a task.
This is a best effort implementation. False positives are permitted. It is permitted for
the service to return Ready
from a poll_ready
call and the next invocation of call
results in an error.
§Notes
.poll_ready()
might be called on different task from actual service call.- In case of chained services,
.poll_ready()
is called for all services at once. - Every
.call()
in chained services enforces readiness.
sourcefn poll_shutdown(&self, cx: &mut Context<'_>) -> Poll<()>
fn poll_shutdown(&self, cx: &mut Context<'_>) -> Poll<()>
Shutdown service.
Returns Ready
when the service is properly shutdowns. This method might be called
after it returns Ready
.
sourcefn map<F, Res>(self, f: F) -> ServiceChain<Map<Self, F, Req, Res>, Req>
fn map<F, Res>(self, f: F) -> ServiceChain<Map<Self, F, Req, Res>, Req>
Map this service’s output to a different type, returning a new service of the resulting type.
This function is similar to the Option::map
or Iterator::map
where it will change
the type of the underlying service.
Note that this function consumes the receiving service and returns a wrapped version of it,
similar to the existing map
methods in the standard library.
sourcefn map_err<F, E>(self, f: F) -> ServiceChain<MapErr<Self, F, E>, Req>
fn map_err<F, E>(self, f: F) -> ServiceChain<MapErr<Self, F, E>, Req>
Map this service’s error to a different error, returning a new service.
This function is similar to the Result::map_err
where it will change the error type of
the underlying service. This is useful for example to ensure that services have the same
error type.
Note that this function consumes the receiving service and returns a wrapped version of it.