moonbitstack/moonzero/syncx

README
U
Balancer
U
BidiCall
U
BidiStreamFactory
U
BidiStreamHandler
U
Breaker
U
Call
U
ClientDeps
U
ClientStreamHandler
U
Clock
U
Conf
U
CorsConf
U
Discovery
U
Endpoint
U
FileEntry
U
Files
U
Group
U
Interceptor
U
Invoker
U
JwtRoute
U
Layer
U
Lease
U
LoadBalancedChannel
U
LogLevel
U
Logger
U
ManualClock
U
Mapper
U
MaxConns
U
MethodTimeout
U
Middleware
U
MiddlewaresConf
U
Once
U
PeriodLimit
U
PeriodResult
U
Pool
U
PrivateKeyConf
U
Promise
U
RedisPeriodLimit
U
RedisTokenLimit
U
Refusal
U
Refused
U
Registrar
U
RequestLog
U
Resolve
U
RestConf
U
RestEngine
U
Ring
U
RouteOptions
U
RoutineGroup
U
RpcChannel
U
RpcClientConf
U
RpcClientMiddlewaresConf
U
RpcCluster
U
RpcGroup
U
RpcHandler
U
RpcMetrics
U
RpcServer
U
RpcServerConf
U
RpcServerMiddlewaresConf
U
Server
U
ServerDeps
U
ServerMetrics
U
ServerStreamHandler
U
Service
U
ServiceConf
U
ServiceGroup
U
Shedder
U
ShutdownCoordinator
U
SignatureConf
U
SignatureInput
U
SignatureKey
U
SignatureVerifier
U
Single
U
StableRunner
U
StatConf
U
TaskRunner
U
Ticket
U
TokenBucket
U
Unavailable
U
Update
U
VerifiedContent
U
View
U
Watcher
V
auth
V
breaker
V
canonical_key
V
chain
V
chunk
V
client_breaker
V
client_duration
V
client_interceptors
V
client_prometheus
V
client_timeout
V
client_tracing
V
content_security
V
cors
V
default_latency_buckets
V
duration_key
V
entry
V
exposition_content_type
V
file_server
V
group_by
V
hold
V
jwt_authorized
V
jwt_sign
V
jwt_verify
V
layer
V
level_name
V
logging
V
logx
V
map_reduce
V
map_reduce_parallel
V
max_conns
V
maxbytes
V
media_type_of
V
message_key
V
metrics
V
metrics_handler
V
mount_metrics
V
not_allowed
V
not_found
V
period_limit
V
period_script
V
plain_text
V
port
V
rate_limit
V
recovery
V
redis_period_limit
V
redis_rate_limit
V
refusal
V
request_id
V
retry
V
retry_with_backoff
V
rpc_port
V
server_auth
V
server_breaker
V
server_interceptors
V
server_prometheus
V
server_recover
V
server_shedding
V
server_stat
V
server_timeout
V
server_tracing
V
set_log_level
V
set_log_sink
V
set_log_writer
V
shedding
V
stamp_key
V
status_code
V
status_name
V
status_of_code
V
structured_logging
V
timed_out
V
timeout
V
token_script
V
too_large
V
too_many
V
trace_clock
V
trace_id_of
V
tracer
V
tracing
V
unavailable
V
verify_signature

moonbitstack/moonzero/syncx does not have a README file

Inflight

type Inflight[T]

One key's call: the waiters' condition, and the answer once it is known.

Once

source

pub struct Once {
  done : Bool
}

A body that runs at most once: the first caller runs it, every later one is a no-op. Guards one-time initialisation, which in a service is a pool, a table or a connection that must not be built twice.

Once::done

source

fn Once::done(self : Once) -> Bool

Whether the body has run.

Once::new

source

fn Once::new() -> Once

A body that has not run yet.

Once::run

source

fn Once::run(self : Once, body : () -> Unit) -> Bool

Run body unless it has already run, and report whether this call ran it.

Pool

source

pub struct Pool[T] {
  make : () -> T
  idle : Array[T]
}

A pool of reusable objects: a caller takes one, uses it, and hands it back. Building is deferred until the first take, so a pool nobody touches costs nothing — which is what makes it usable for a buffer or a connection that only some code paths need.

Pool::get

source

fn[T] Pool::get(self : Pool[T]) -> T

Take an object: an idle one when there is one, a fresh one otherwise.

Pool::idle

source

fn[T] Pool::idle(self : Pool[T]) -> Int

How many objects are idle.

Pool::new

source

fn[T] Pool::new(make : () -> T) -> Pool[T]

A pool that builds its objects with make.

Pool::put

source

fn[T] Pool::put(self : Pool[T], item : T) -> Unit

Hand an object back. The pool keeps every one it is given, so a caller that wants a ceiling on retained objects puts one on its own side.

Single

source

pub struct Single[T] {
  inflight : Map[String, Inflight[T]]
}

One call in flight per key: a second caller asking for the same key while the first is still running waits for that answer instead of issuing a second call, and every one of them receives the same result — or the same failure.

This is the primitive a cache-aside store is built on: a miss under load must put one query on the database, not one per request.

Single::len

source

fn[T] Single::len(self : Single[T]) -> Int

The number of calls in flight, which is also the number of keys being shared.

Single::new

source

fn[T] Single::new() -> Single[T]

An empty group: no call in flight.

Single::run

source

async fn[T] Single::run(self : Single[T], key : String, body : async () -> T) -> T

Run body for key, or wait for the call already running for it.

The caller that runs the body is the one that answers everyone; a body that raises answers them all with the same error, so a failing lookup is not retried once per waiter.

Source Files

Install

Download zip

A service framework for MoonBit.

go-zeromicroservicemicroservicesservice-frameworkframeworkrestrpczrpcgrpcmiddlewarecircuit-breakerrate-limitload-sheddingload-balancingservice-discoverymetricstracingloggingconfigyamljwtsignaturecorsgzipssefile-serverp2csingleflightcacheredissqlprometheusopentelemetrygraceful-shutdownmoonapiservermoonbit

moonbitstack

Apache-2.0