pub fn return_trigger(
flow: &Flow,
evidence: &[StageEvidence],
) -> Option<(usize, u32)>Expand description
The failure that most recently sent the walk backwards, as the (stage index, attempt) key of its evidence row — the row a re-entered stage is
being re-run because of.
Every backward edge leaves a Fail row on the stage that owns it, so the
last such row in the log is the jump the walk is still inside. Reading it
from the persisted log rather than from a live counter is what makes a
re-run prompt reproducible: a resumed run derives the same trigger, and so
composes the same prompt, as the daemon that first took the jump.