Struct InMemoryKernelJournal 

pub struct InMemoryKernelJournal { /* private fields */ }

Single-process dev/test implementation (spec Task 8b, criterion 2).

CAS is genuinely atomic here — every check-then-mutate below happens while the single Mutex is held, and none of those critical sections contains an .await, so the compiler itself guarantees no task interleaves inside one — but that atomicity ends at the process boundary. Two processes sharing “the same” journal do not exist: each has its own map. Production hosts must supply a KernelJournal whose CAS is a real storage-layer primitive (spec §9.1); FileKernelJournal is the reference for that.

Implementations

impl InMemoryKernelJournal

pub fn new() -> Self

Trait Implementations

impl Default for InMemoryKernelJournal

fn default() -> Self

Returns the “default value” for a type.

impl KernelJournal for InMemoryKernelJournal

fn stage_outbound_envelope<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, operation_id: &'life1 str, envelope_json: &'life2 str, ) -> Pin<Box<dyn Future<Output = JournalResult<()>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

Persist the byte-identical canonical input envelope before attempting its record append.

fn read_outbound_envelope<'life0, 'life1, 'async_trait>( &'life0 self, operation_id: &'life1 str, ) -> Pin<Box<dyn Future<Output = JournalResult<Option<String>>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Return the staged outbound envelope, if an append-before crash left one behind.

fn clear_outbound_envelope<'life0, 'life1, 'async_trait>( &'life0 self, operation_id: &'life1 str, ) -> Pin<Box<dyn Future<Output = JournalResult<()>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Clear a staged outbound envelope after its input is durably owned or rejected.

fn compare_and_append<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, operation_id: &'life1 str, expected_head: Option<&'life2 str>, record: JournalRecordInput, ) -> Pin<Box<dyn Future<Output = JournalResult<JournalAppendReceipt>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

Atomically append record iff the operation’s head is exactly expected_head.

fn head<'life0, 'life1, 'async_trait>( &'life0 self, operation_id: &'life1 str, ) -> Pin<Box<dyn Future<Output = JournalResult<Option<JournalHead>>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

The current head, or None when the operation has no records and no pruned anchor.

fn read_from<'life0, 'life1, 'async_trait>( &'life0 self, operation_id: &'life1 str, from_step_seq: u64, ) -> Pin<Box<dyn Future<Output = JournalResult<Vec<JournalEntry>>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Records with step_seq >= from_step_seq, in chain order. 0 returns the whole retained chain (Rust has no default arguments; node’s optional parameter defaults to the same).

fn records_after<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, operation_id: &'life1 str, after_head: Option<&'life2 str>, ) -> Pin<Box<dyn Future<Output = JournalResult<Vec<JournalEntry>>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

Records strictly after the record whose digest is after_head — the digest-anchored cursor §9.1 names records_after(operation_id, checkpoint_head). None returns everything retained.

fn compare_and_install_checkpoint<'life0, 'life1, 'life2, 'life3, 'async_trait>( &'life0 self, operation_id: &'life1 str, previous_checkpoint_id: Option<&'life2 str>, covered_head: &'life3 str, checkpoint: CheckpointCandidate, ) -> Pin<Box<dyn Future<Output = JournalResult<InstalledCheckpoint>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait, 'life3: 'async_trait,

Atomically install checkpoint iff the operation’s checkpoint pointer is exactly previous_checkpoint_id, and covered_head is the record digest at checkpoint.through_step_seq.

fn latest_checkpoint<'life0, 'life1, 'async_trait>( &'life0 self, operation_id: &'life1 str, ) -> Pin<Box<dyn Future<Output = JournalResult<Option<InstalledCheckpoint>>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

The highest-ordinal installed checkpoint, acknowledged or not.

fn ack_checkpoint<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, operation_id: &'life1 str, checkpoint_id: &'life2 str, ) -> Pin<Box<dyn Future<Output = JournalResult<InstalledCheckpoint>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

Record the durable acknowledgement that opens the prefix-reclamation boundary. Idempotent.

fn prune_acked_prefix<'life0, 'life1, 'async_trait>( &'life0 self, operation_id: &'life1 str, ) -> Pin<Box<dyn Future<Output = JournalResult<JournalPruneReceipt>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Reclaim the record prefix covered by the latest acknowledged checkpoint. A no-op while no checkpoint is acknowledged. The pruned boundary is retained as an anchor so Self::head and the next CAS still resolve on a fully-pruned chain (Task 8b correction (d)).