Struct ArmOutcome 

pub struct ArmOutcome {
    pub passed: bool,
    pub iterations: u32,
    pub cost_usd: f64,
    pub wall_ms: u64,
    pub error: Option<String>,
    pub transcript_path: Option<String>,
    pub infra_failed: bool,
}

A single arm’s result on a single task, normalized so both the native LoopOutcome and the external CLI’s result map onto one shape.

Fields

passed: bool

Every contract check passed (the ground-truth win condition).

iterations: u32

Contract-evaluation rounds actually executed.

cost_usd: f64

USD spent on inference for this arm’s run (0.0 when unknown — the native loop does not always meter; the external CLI reports total_cost_usd).

wall_ms: u64

Wall-clock for the arm’s run.

error: Option<String>

Terminal error (cancellation, repeated inference failure). Distinct from infra_failed: an arm can error late (task-attributable) after really running.

transcript_path: Option<String>

Where the arm’s transcript/event stream landed, for Slice 3 attribution.

infra_failed: bool

The arm never actually attempted the task — worktree/setup/CLI-launch/ transport failure before any real work. Excluded from the honest denominator (ALE’s infra-vs-task split); NOT a scored task-0.

Implementations

impl ArmOutcome

pub fn infra(reason: impl Into<String>, wall_ms: u64) -> Self

An infra failure: the arm couldn’t be attempted. Kept out of the scored denominator.

pub fn timeout_loss(reason: impl Into<String>, wall_ms: u64) -> Self

An arm that ran but was stopped by the harness’s own wall bound.

A scored loss, deliberately — not ArmOutcome::infra. The infra exclusion means “the arm could not be attempted”; an arm that ran for its entire wall budget was attempted, and very hard. Excluding it would drop precisely the SLOW tasks from the denominator, which is the same upward bias coder_ab::INFRA_MARKERS documents at length and warns against — it excluded “precisely the dependency-trouble tasks, i.e. the hard ones”.

The asymmetry made it worse: the two arms carry different wall bounds (900s native, 300s external), so excluding on timeout gave a paired design two different exclusion rates feeding one denominator.

pub fn scorable(&self) -> bool

Whether this outcome counts toward the scored denominator.

Trait Implementations

impl Clone for ArmOutcome

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ArmOutcome

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for ArmOutcome

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for ArmOutcome

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

Equality operator ==.

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

Inequality operator !=.

impl Serialize for ArmOutcome

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 ArmOutcome