pub fn classify_run_outcome(
success: bool,
records_empty: bool,
has_failed_cell: bool,
scheduler: Option<&str>,
exit_code: Option<i32>,
) -> RunOutcomeExpand description
Decide the cargo ktstr verifier run outcome from nextest’s exit
success, whether any per-cell records were produced, whether any of
those records is a FAILED cell (already shown as a ✗ in the printed
grid), and the optional --scheduler filter.
The dispatcher runs nextest with --no-tests=pass, so a run that
selects zero verifier cells exits 0 (success) with empty records
rather than nextest’s generic “no tests to run” (exit 4). That lets
the dispatcher diagnose the empty case itself instead of surfacing a
cryptic nextest exit:
--scheduler <NAME>set + no records: the name is not a declared scheduler, or its declared topology constraints / verifier-only exclusions reject every selected preset.- no
--scheduler+ no records: nodeclare_scheduler!is linked into a test binary the sweep sees, or every declared scheduler rejects all selected presets through those declaration-level gates.
A genuine build/exec failure still fails nextest (exit non-zero, which
--no-tests=pass does not mask). When that nonzero exit is a REAL
cell failure — has_failed_cell true and nextest handed back a
numeric exit_code — the printed grid already shows the failing
cell(s), so the outcome is RunOutcome::SilentExit (carry nextest’s
code, emit no stderr line). Every other nonzero case — a build /
internal failure that produced no failed cell, or a signal exit with
no numeric code — stays RunOutcome::Failed with the descriptive
message, as does each empty-record diagnostic.