Struct LanceCache 

pub struct LanceCache { /* private fields */ }

Typed cache wrapper that handles key construction and type safety.

Internally delegates to a CacheBackend. The default backend is MokaCacheBackend; pass a custom backend via LanceCache::with_backend.

Implementations

impl LanceCache

pub fn with_capacity(capacity: usize) -> Self

pub fn with_backend(backend: Arc<dyn CacheBackend>) -> Self

Create a cache backed by a custom CacheBackend.

pub fn no_cache() -> Self

pub fn with_key_prefix(&self, prefix: &str) -> Self

Derive a child namespace for all keys in the returned cache handle.

Each call adds one framed hierarchy segment. Consequently, cache.with_key_prefix("a").with_key_prefix("b") is deliberately distinct from cache.with_key_prefix("a/b").

pub async fn size(&self) -> usize

pub fn approx_size(&self) -> usize

pub async fn size_bytes(&self) -> usize

pub async fn stats(&self) -> CacheStats

pub async fn clear(&self)

pub async fn insert_with_key<K>( &self, cache_key: &K, metadata: Arc<K::ValueType>, )

where K: CacheKey, K::ValueType: DeepSizeOf + Send + Sync + 'static,

pub async fn get_with_key<K>(&self, cache_key: &K) -> Option<Arc<K::ValueType>>

where K: CacheKey, K::ValueType: DeepSizeOf + Send + Sync + 'static,

pub async fn get_or_insert_with_key<K, F, Fut>( &self, cache_key: K, loader: F, ) -> Result<Arc<K::ValueType>>

where K: CacheKey, K::ValueType: DeepSizeOf + Send + Sync + 'static, F: FnOnce() -> Fut + Send, Fut: Future<Output = Result<K::ValueType>> + Send,

pub async fn get_or_insert_with_key_hit<K, F, Fut>( &self, cache_key: K, loader: F, ) -> Result<(Arc<K::ValueType>, bool)>

where K: CacheKey, K::ValueType: DeepSizeOf + Send + Sync + 'static, F: FnOnce() -> Fut + Send, Fut: Future<Output = Result<K::ValueType>> + Send,

Same as get_or_insert_with_key, but also returns a boolean indicating whether the loader was skipped for this call.

• true means this call did not execute the loader. That covers both a true cache hit on an already-populated entry and a coalesced concurrent load where an in-flight loader started by a different caller produced the value.
• false means the loader ran on this call (a real cache miss).

Callers that want strict “served from cache” semantics should treat coalesced loads as misses; the current backend does not distinguish the two cases. Prefer this over rolling a caller-side Arc<AtomicBool> when the caller needs per-query hit/miss counters — the backend already tracks this bit internally and this method just exposes it.

pub async fn insert_unsized_with_key<K>( &self, cache_key: &K, metadata: Arc<K::ValueType>, )

where K: UnsizedCacheKey, K::ValueType: DeepSizeOf + Send + Sync + 'static,

pub async fn get_unsized_with_key<K>( &self, cache_key: &K, ) -> Option<Arc<K::ValueType>>

where K: UnsizedCacheKey, K::ValueType: DeepSizeOf + Send + Sync + 'static,

Trait Implementations

impl Clone for LanceCache

fn clone(&self) -> LanceCache

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for LanceCache

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl DeepSizeOf for LanceCache

fn deep_size_of_children(&self, _: &mut Context) -> usize

fn deep_size_of(&self) -> usize