Struct SecondOpinionInfo 

pub struct SecondOpinionInfo {
    pub tried: Vec<String>,
    pub promoted_by: Option<String>,
    pub base_status: String,
    pub base_iteration_count: usize,
    pub rung_iteration_counts: Vec<usize>,
    pub total_iteration_count: usize,
}

The second-opinion ladder’s record: which rungs ran, which (if any) was promoted, and what the base solve did before any of it (gh #850).

Why the base solve’s numbers are the point

On a promotion the reported status and statistics.iteration_count both become the promoted rung’s, and nothing else in the report says the base solver failed. That makes a lost solve indistinguishable from a faster one, and it is worse than a gap in the evidence: it produces positive evidence for the wrong conclusion.

The case that exposed it is square_flowsheet_resto, where v0.10.0’s base solver converged in 116 iterations and HEAD’s does not converge at all — RestorationFailed at 131 — with the answer coming from a ladder rung (start_point_perturbation=1e-2) added in the same release, which promotes at 54. scripts/sweep-fixtures.sh read that as 116 -> 54, a 2× win.

The cost is understated on top of that: statistics.iteration_count is the promoted rung’s alone, so the fixture’s true cost is 131 + 54, 3.4× what the report says. Self::total_iteration_count is the honest number.

This is the same shape of invisibility the sweep’s engine column was added to close, and CLAUDE.md’s rule applies verbatim: a line whose only moving field is “solved directly” → “promoted from a rung” is a trajectory change, and is as reportable as a moved iteration count.

Fields

tried: Vec<String>

Rung labels actually run, in order.

promoted_by: Option<String>

The rung whose re-solve was promoted, or None when the original verdict survived every rung and shipped unchanged.

base_status: String

The verdict the ladder was opened on, before any rung ran.

base_iteration_count: usize

Iterations the base solve spent. Not included in statistics.iteration_count when a rung was promoted.

rung_iteration_counts: Vec<usize>

Iterations each rung in tried spent, in the same order.

total_iteration_count: usize

Base plus every rung: what the solve actually cost.

Trait Implementations

impl Clone for SecondOpinionInfo

fn clone(&self) -> SecondOpinionInfo

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for SecondOpinionInfo

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for SecondOpinionInfo

fn deserialize<__D>( __deserializer: __D, ) -> Result<SecondOpinionInfo, <__D as Deserializer<'de>>::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for SecondOpinionInfo

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

where __S: Serializer,

Serialize this value into the given Serde serializer.