Struct Cache 

pub struct Cache<'a> { /* private fields */ }

Names what a graph produces, and keeps it in a store.

Implementations

impl<'a> Cache<'a>

pub fn over(store: &'a dyn Store) -> Self

The cache kept in this store.

Trait Implementations

impl Keeper for Cache<'_>

fn present(&self, keys: &[&Key]) -> Result<Vec<bool>, KeeperError>

One scan and no fetches, which is the whole reason the engine asks this instead of reading: what it wants to know is whether it can skip the node underneath, and the bytes of the answer are somebody else’s business — often nobody’s.

fn key_of(&self, value: &Value) -> Option<Key>

The key of a value by its content, which only a root needs: from there down, keys come from keys. None if the value cannot leave this process, which is not a failure — nothing below it is cached either.

fn combine(&self, parts: &[&str]) -> Key

One key out of the ingredients of a recipe, in the order given. The parts have to stay apart: run together, ["ab", "c"] and ["a", "bc"] would name the same thing.

fn recall(&self, keys: &[&Key]) -> Result<Vec<Option<Kept>>, KeeperError>

What is kept under each of these, in the order they were asked. In batch form from the first day: against a remote store, one question per item is one round trip per item.

fn keep( &self, key: &Key, value: &Value, meta: &[(&str, &str)], ) -> Result<(), KeeperError>

Keeps this, with what should be remembered beside it — the fingerprint of the code that produced it, above all, which is not in the key and is what a hit gets compared against.