Struct DiscreteFileStoreNonVolatile 

pub struct DiscreteFileStoreNonVolatile<Key, Value, Serde>
where
    Key: Clone + Eq + Hash + Serialize + for<'a> Deserialize<'a>,
    Value: Serialize + for<'a> Deserialize<'a>,
    Serde: DiscreteFileSerializerDeserializer<Value>,
{ /* private fields */ }

Before using, strongly consider using the volatile version. Do you really need this cache to rehydrate without hitting the source of record? You are sacrificing reboot to clear corruption and now must consider N vs N+1 schema issues when downgrading or upgrading your application.

Implementations

impl<Key, Value, Serde> DiscreteFileStoreNonVolatile<Key, Record<Key, Value>, Serde>

where Key: Clone + Eq + Hash + Serialize + for<'a> Deserialize<'a>, Value: Serialize + for<'a> Deserialize<'a>, Serde: DiscreteFileSerializerDeserializer<Record<Key, Value>>,

pub fn new(cache_directory: PathBuf) -> Self

Trait Implementations

impl<Key, Value, Serde> CacheStoreStrategy<Key, Value> for DiscreteFileStoreNonVolatile<Key, Record<Key, Value>, Serde>

where Key: Clone + Eq + Hash + Serialize + for<'a> Deserialize<'a>, Value: Serialize + for<'a> Deserialize<'a>, Serde: DiscreteFileSerializerDeserializer<Record<Key, Value>>,

fn get(&self, key: &Key) -> Option<Value>

fn peek(&self, key: &Key) -> Option<Value>

A platform read of a value. When replacement strategies are used (e.g. LRU) reads have side effects that update internal tracking. If hydration requires inspecting the current state, these reads will skew tracking. Peek allows you to inspect state without side effects, it signals to any layer that a platform read has occurred that should be ignored for usage tracking purposes.

fn put(&mut self, key: &Key, value: Value)

fn delete(&mut self, key: &Key) -> bool

fn flush(&mut self)

fn get_keys(&self) -> KeyIterator<'_, Key>

fn contains(&self, key: &Key) -> bool