Struct CompactionPolicy 

pub struct CompactionPolicy {
    pub auto_enabled: bool,
    pub auto_threshold_percent: u8,
    pub tail_turns: usize,
    pub tail_token_budget: usize,
    pub tool_output_max_chars: usize,
    pub summary_max_tokens: usize,
    pub summarizer_input_token_budget: usize,
    pub min_response_reserve_tokens: usize,
    pub max_response_reserve_tokens: usize,
}

Fields

auto_enabled: bool

auto_threshold_percent: u8

tail_turns: usize

tail_token_budget: usize

tool_output_max_chars: usize

summary_max_tokens: usize

summarizer_input_token_budget: usize

min_response_reserve_tokens: usize

max_response_reserve_tokens: usize

Implementations

impl CompactionPolicy

pub fn summary_output_tokens(self, window: Option<usize>) -> usize

The summarizer’s output cap for a model whose window is window.

Unknown window keeps the flat summary_max_tokens (the remote-provider case — nothing better to go on). A known window caps the summary at 1/SUMMARY_OUTPUT_WINDOW_SHARE of it, so the input budget always has room left; large windows are unaffected because summary_max_tokens stays the ceiling.

pub fn summary_input_budget(self, window: Option<usize>) -> usize

Total input budget (fixed scaffold + history excerpt) for the summarizer, given the model’s window.

Monotonic in the window, which the previous computation was NOT: it subtracted a flat output cap and then treated a non-positive result as “window unknown”, falling back to the most permissive budget. An 8k model therefore got a 64k input budget while a 9k model got 1k — shrinking the window made the request bigger, and 8k-and-below could never compact.

pub fn response_reserve(self, request: &ChatRequest) -> usize

Window room to hold back for the model’s response when sizing compaction. Decoupled from the on-wire output cap: an explicit user cap is the best reserve estimate, but AUTO (max_tokens == 0) reserves the baseline plus the reasoning headroom the level implies — a High/Max turn needs more room before the window counts as “full”.

Trait Implementations

impl Clone for CompactionPolicy

fn clone(&self) -> CompactionPolicy

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for CompactionPolicy

impl Debug for CompactionPolicy

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

Formats the value using the given formatter.

impl Default for CompactionPolicy

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for CompactionPolicy

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Eq for CompactionPolicy

impl PartialEq for CompactionPolicy

fn eq(&self, other: &CompactionPolicy) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl Serialize for CompactionPolicy

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for CompactionPolicy