Struct CacheConfig 

pub struct CacheConfig {
    pub memory_bytes: usize,
    pub disk: Option<DiskCache>,
    pub memory_auto: Option<MemoryPolicy>,
    pub write_queue_bytes: usize,
    pub writers: Option<usize>,
}

Cache configuration for an Engine.

Fields

memory_bytes: usize

In-memory LRU capacity for inflated entries (0 disables it).

disk: Option<DiskCache>

On-disk cache shared across processes.

memory_auto: Option<MemoryPolicy>

Treat memory_bytes as a starting point and keep resizing the LRU to this process’s share of one budget for every resident host the user runs (MemoryPolicy::share), instead of a fixed capacity.

write_queue_bytes: usize

Heap queued disk writes may hold before new ones are deferred ([WriteJob::defer]); a memory cache’s budget lowers it further.

writers: Option<usize>

Disk writer threads (None: CacheConfig::auto_writers).

Implementations

impl CacheConfig

pub const WRITE_QUEUE: usize

Default of CacheConfig::write_queue_bytes. A cold multi-APK sweep inflates far faster than the writers persist, and each queued job keeps its DEX bytes or tables alive until written; without a bound the heap grows with the backlog (tens of GB over a phone’s APKs).

pub const MAX_AUTO_WRITERS: usize = 4

Most automatic writer threads: writing is I/O, but a deferred L1 write inflates its entry again first.

pub fn disk_default() -> CacheConfig

Disk cache in the default directory, no in-memory cache (CLI default).

pub fn resident(memory_bytes: usize) -> CacheConfig

Disk cache plus an in-memory LRU of a fixed capacity.

pub fn auto_writers(cpus: usize) -> usize

One writer per 8 CPUs, 1..=CacheConfig::MAX_AUTO_WRITERS.

Trait Implementations

impl Default for CacheConfig

fn default() -> CacheConfig

Returns the “default value” for a type.