kernel/me.ahoo.wow.kernel/KernelOptions

KernelOptions

class KernelOptions @InternalWowApi constructor(val maxAttempts: Int = DEFAULT_MAX_ATTEMPTS, val verifyDecidePurity: Boolean = false, val reservationTtl: Duration = DEFAULT_RESERVATION_TTL, val maxEarlyWait: Duration = DEFAULT_MAX_EARLY_WAIT, val recentRequests: Int = Kernel.DEFAULT_RECENT_REQUESTS, val maxReplay: Int = DEFAULT_MAX_REPLAY)

The tuning of a Kernel, set in code, not by configuration keys:

maxAttempts: the times a command is decided again after a version conflict before it is answered as contended;
verifyDecidePurity: whether every decide function is checked not to change the state it was given (the test DSL turns it on; it costs two serializations of the state per command);
reservationTtl: how long a pending reservation lives unless the append confirms it first;
maxEarlyWait: how long a scheduled command due later by the kernel's clock is held for its slot before it fails as not due, to be sent again;
recentRequests: the requests each cached aggregate keeps of those it committed last, so that most commands skip looking their request id up (0 looks every one up);
maxReplay: the most streams a read of a state at a version folds (Kernel.stateAt): a state further from its snapshot, or from version 1 without one, is refused with ReplayLimitExceededException, which names the aggregate, instead of replaying a long history on one read.

Fields may be added: construct it with named arguments; the constructor is internal to Wow.

Constructors

KernelOptions

@InternalWowApi

constructor(maxAttempts: Int = DEFAULT_MAX_ATTEMPTS, verifyDecidePurity: Boolean = false, reservationTtl: Duration = DEFAULT_RESERVATION_TTL, maxEarlyWait: Duration = DEFAULT_MAX_EARLY_WAIT, recentRequests: Int = Kernel.DEFAULT_RECENT_REQUESTS, maxReplay: Int = DEFAULT_MAX_REPLAY)