Struct LaunchOverrides 

pub struct LaunchOverrides {
    pub options: ServerConfigOptions,
    pub cache_ram: Option<CacheRamSetting>,
}

Per-call launch overrides layered on a runtime’s standing configuration.

A runtime is normally built once with a standing template — the proxy’s cache settings, say — and then shared, so that only one llama-server runs at a time. This is how an individual caller contributes launch options on top of that template without needing a manager of its own.

Default means “no opinion”: every field falls through to the template.

Fields

options: ServerConfigOptions

Explicit options merged over the runtime’s template, Some fields winning — see ServerConfigOptions::overlay.

cache_ram: Option<CacheRamSetting>

How to size llama-server’s host-RAM prompt cache for this launch.

Separate from Self::options because it is resolved at spawn against live system RAM and the model’s KV footprint, not carried as a flag. None defers to the runtime’s own setting.

Trait Implementations

impl Clone for LaunchOverrides

fn clone(&self) -> LaunchOverrides

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for LaunchOverrides

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

Formats the value using the given formatter.

impl Default for LaunchOverrides

fn default() -> LaunchOverrides

Returns the “default value” for a type.