Struct WgslModuleCache 

pub struct WgslModuleCache { /* private fields */ }

Memoizes runtime-compiled WGSL modules for one device.

Runtime-assembled shaders (fused filter chains, specialized spatial passes, user fragment surfaces) are compiled from a fully substituted source string, so that string is the specialization key: two callers that produce byte-identical WGSL must share one wgpu::ShaderModule.

One cache belongs to one device. Sharing a cache across devices would hand out modules the other device cannot use, so hosts own a cache next to the device they created it for.

The map is behind a Mutex because this is pure memoization with no cross-thread protocol to model as a channel: lookups happen during pipeline setup, and a hit costs one hash of the source.

Implementations

impl WgslModuleCache

pub fn new() -> WgslModuleCache

Creates an empty cache for one device.

pub fn module( &self, device: &Device, label: Option<&str>, source: &str, ) -> ShaderModule

Returns the module compiled from source, compiling it on first use.

Panics

Panics when a previous caller panicked while holding the cache lock.

pub fn len(&self) -> usize

Number of distinct sources compiled through this cache.

Panics

Panics when a previous caller panicked while holding the cache lock.

pub fn is_empty(&self) -> bool

Whether this cache has compiled anything yet.

Trait Implementations

impl Debug for WgslModuleCache

fn fmt(&self, formatter: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl Default for WgslModuleCache

fn default() -> WgslModuleCache

Returns the “default value” for a type.