Struct PreparedStatementCache 

pub struct PreparedStatementCache { /* private fields */ }

A cache for prepared statements.

Tracks which SQL strings have been prepared so we emit a trace! for hits vs. misses. Eviction is true LRU via lru::LruCache — when the cache reaches max_size the least-recently-used entry is dropped on the next insert.

The cache is keyed on the SQL string; the actual Statement is fetched from client.prepare_cached on every call (deadpool reuses its own per-connection cache).

Implementations

impl PreparedStatementCache

pub fn new(max_size: usize) -> Self

Create a new statement cache with the given maximum size.

max_size of 0 is treated as 1 to satisfy NonZeroUsize.

pub async fn get_or_prepare( &self, client: &Object, sql: &str, ) -> PgResult<Statement>

Get or prepare a statement for the given SQL.

pub async fn get_or_prepare_in_txn<'a>( &self, txn: &Transaction<'a>, sql: &str, ) -> PgResult<Statement>

Get or prepare a statement within a transaction.

pub fn clear(&self)

Clear all cached statements.

pub fn evict(&self, sql: &str) -> bool

Forget a single cached SQL string, so the next get_or_prepare for it records a miss and re-prepares.

Used to recover from PostgreSQL 0A000 "cached plan must not change result type": DDL altered a referenced table’s result columns while a prepared statement for this SQL was cached on the connection. Dropping the key here, combined with a fresh (uncached) prepare on the retry, re-plans against the current schema. Returns whether the key was present.

pub fn len(&self) -> usize

Get the number of cached statement keys.

pub fn is_empty(&self) -> bool

Check if the cache is empty.

pub fn max_size(&self) -> usize

Get the maximum cache size.