Struct Journal 

pub struct Journal<B> { /* private fields */ }

The single state-machine implementation shared by every backend.

Implementations

impl<B: JournalBackend> Journal<B>

pub fn new(backend: B) -> Self

Wraps a backend. Durable production backends must report write safety.

pub fn acquire_lease(&self) -> Result<Lease, JournalError>

Obtains a new fencing generation, invalidating every older lease.

pub fn head(&self) -> Result<Option<JournalHead>, JournalError>

Returns the current verified head.

pub fn publish( &self, lease: &Lease, expected: Option<&JournalHead>, objects: Vec<JournalObject>, entries: Vec<JournalEntry>, ) -> Result<JournalHead, JournalError>

Atomically publishes objects and an ordered entry batch under one fence.

Exact redelivery is a no-op. Any conflicting delivery, missing/corrupt payload, stale fence, sequence gap, or wrong predecessor is rejected.

pub fn verify(&self) -> Result<Verification, JournalError>

Reads and verifies the complete journal closure.

pub fn replay(&self) -> Result<Replay, JournalError>

Replays verified entries in sequence order.

pub fn verified_snapshot(&self) -> Result<VerifiedSnapshot, JournalError>

Reads and verifies one internally consistent semantic snapshot.

Consumers that must reduce entries together with their payload values use this method so both come from the same backend read. The returned value exposes no physical storage identity or mutable backend state.

pub fn table_projection(&self) -> Result<TableProjection, JournalError>

Creates a detached, read-only table projection.

pub fn dir_projection(&self) -> Result<DirProjection, JournalError>

Creates a detached, read-only directory projection.

pub fn entry( sequence: u64, previous: Option<ContentId>, kind: Symbol, payloads: Vec<ContentId>, ) -> JournalEntry

Convenience constructor for an entry with an open kind symbol.