Struct QueryPlanCache 

pub struct QueryPlanCache { /* private fields */ }

A simple LRU query-plan cache with O(n) eviction.

This is deliberately kept dependency-free (no lru crate, no SciRS2). The access order vector has the most-recently-used entry at the front (index 0) and the least-recently-used at the back.

For the high-throughput LRU cache backed by the lru crate, see LruQueryPlanCache.

Implementations

impl QueryPlanCache

pub fn new(max_size: usize) -> Self

Create a new cache that can hold at most max_size plans.

If max_size is 0 it is treated as 1 to avoid degenerate behaviour.

pub fn get(&mut self, query_hash: u64) -> Option<&QueryPlan>

Retrieve the plan for query_hash.

On a hit the entry is moved to the front of the access-order list and stats.hits is incremented. On a miss stats.misses is incremented.

pub fn insert(&mut self, plan: QueryPlan)

Insert a plan into the cache.

If the cache is at capacity, evict_lru is called first. The new entry is placed at the front of the access-order list.

pub fn evict_lru(&mut self) -> Option<QueryPlan>

Remove and return the least-recently-used plan (the one at the back of the access-order list). Increments stats.evictions.

Returns None if the cache is empty.

pub fn invalidate(&mut self, query_hash: u64) -> bool

Remove a specific plan by hash. Returns true if it existed.

pub fn clear(&mut self)

Remove all entries and reset size to zero (other statistics are kept).

pub fn stats(&self) -> &PlanCacheStats

Return a snapshot of the current statistics.

pub fn hit_rate(&self) -> f64

Cache hit rate: hits / (hits + misses), or 0.0 if no operations have been performed yet.

pub fn len(&self) -> usize

Current number of cached plans.

pub fn is_empty(&self) -> bool

Return true if the cache contains no plans.