| Safe Haskell | Safe-Inferred |
|---|---|
| Language | Haskell2010 |
Test.Consensus.PeerSimulator.ScheduledServer
Description
This module contains code that is generic to any “scheduled server” (think scheduled ChainSync or BlockFetch server). A scheduled server keeps track of the current state of a point schedule and wakes up when new ticks arise. It processes as many messages there are via its domain-specific handlers; once there is nothing new to process, or what needs to process requires a different state of the point schedule, the scheduled server goes back to sleep, awaiting another tick.
Synopsis
- data ScheduledServer m state blk = ScheduledServer {
- ssPeerId ∷ PeerId
- ssCurrentState ∷ STM m (Maybe state)
- ssTickStarted ∷ STM m ()
- ssCommonTracer ∷ Tracer m (TraceScheduledServerHandlerEvent state blk)
- awaitOnlineState ∷ IOLike m ⇒ ScheduledServer m state blk → m state
- ensureCurrentState ∷ IOLike m ⇒ ScheduledServer m state blk → m state
- runHandler ∷ IOLike m ⇒ ScheduledServer m state blk → String → (state → STM m (Maybe msg, [traceMsg])) → Tracer m traceMsg → (msg → m h) → m h
- runHandlerWithTrace ∷ IOLike m ⇒ Tracer m traceMsg → STM m (state, [traceMsg]) → m state
Documentation
data ScheduledServer m state blk Source #
Constructors
| ScheduledServer | |
Fields
| |
awaitOnlineState ∷ IOLike m ⇒ ScheduledServer m state blk → m state Source #
Block until the peer simulator has updated the concurrency primitive that
indicates that it's this peer's server's turn in the point schedule.
If the new state is Nothing, the point schedule has declared this peer as
offline for the current tick, so it will not resume operation and wait for
the next update.
ensureCurrentState ∷ IOLike m ⇒ ScheduledServer m state blk → m state Source #
Fetch the current state from the STM action, and if it is Nothing,
wait for the next tick to be triggered in awaitOnlineState.
Since processing of a tick always ends when the handler finishes after serving the last point, this function is only relevant for the initial state update.
runHandler ∷ IOLike m ⇒ ScheduledServer m state blk → String → (state → STM m (Maybe msg, [traceMsg])) → Tracer m traceMsg → (msg → m h) → m h Source #
Run a peer server's message handler by fetching state from the scheduler's STM interface.
The handler is an STM action that returns a protocol result and log messages.
If the result is Nothing, the server's activity for the current tick is complete
and we listen for the scheduler's signal to start the next tick, which we continue without
updating the protocol handler (in restart).
Otherwise, the result is passed to dispatchMessage, which produces a native protocol handler
message with the server's continuation in it.
runHandlerWithTrace ∷ IOLike m ⇒ Tracer m traceMsg → STM m (state, [traceMsg]) → m state Source #
Handler functions are STM actions for the usual race condition reasons, which means that they cannot emit trace messages.
For that reason, we allow them to return their messages alongside the protocol result and emit them here.