Struct ResponseCache 

pub struct ResponseCache { /* private fields */ }

Executor-level cache for projected GraphQL responses.

Stores the final serialized response keyed by (query_hash, security_hash). On hit, the entire projection + RBAC + serialization pipeline is skipped.

Thread Safety

moka::sync::Cache is Send + Sync with lock-free reads. The view reverse index uses DashMap (fine-grained shard locking). There is no global mutex on the read path.

Implementations

impl ResponseCache

pub fn new(config: ResponseCacheConfig) -> Self

Create a new response cache from configuration.

pub const fn is_enabled(&self) -> bool

Whether the response cache is enabled.

pub fn get( &self, query_key: u64, security_hash: u64, ) -> Result<Option<Arc<Value>>>

Look up a cached response.

Errors

This method is infallible with the moka backend and always returns Ok.

pub fn invalidation_generation(&self) -> u64

Snapshot of the invalidation generation, for fencing a cache write (#1079).

Take this before the work whose result is being cached, and pass it to put.

pub fn put( &self, query_key: u64, security_hash: u64, response: Arc<Value>, accessed_views: Vec<String>, fence: Option<u64>, ) -> Result<()>

Store a response in the cache.

fence is a snapshot from invalidation_generation taken before the work that produced response. If an invalidation landed since, the entry is discarded rather than stored — the caller keeps its response, it is simply not cached. None stores unconditionally.

Errors

This method is infallible with the moka backend and always returns Ok.

pub fn invalidate_views(&self, views: &[ViewName]) -> Result<u64>

Invalidate all entries that access any of the given views.

Uses the O(k) reverse index — no full-cache scan.

Errors

This method is infallible with the moka backend and always returns Ok.

pub fn metrics(&self) -> (u64, u64)

Get cache hit/miss counts.