Enum Severity 

pub enum Severity {
    Off,
    Suggest,
    Info,
    Warn,
    Error,
    Fix,
}

Rule severity level. Configurable per rule in .marque.toml.

Ordering

The derived Ord is Off < Suggest < Info < Warn < Error < Fix. The ordering is exposed for consumers that want to compare severities (e.g., “is this at least Error?”) but the config loader does not use it as a merge operator today. Suggest sits between Off and Info because it is the lightest firing-but-non-actionable channel — quieter than Info (which has no candidate replacement attached) and louder than Off (which is non-firing entirely).

Exit-code semantics

marque check maps severities to exit codes as follows:

Severity counts present	Exit code
Error or Fix	1 (EX_DIAG_ERROR)
Warn only	2 (EX_DIAG_WARN)
Info / Suggest only, none	0 (EX_OK)

Info and Suggest are the only severities whose diagnostics are emitted and keep the exit code at zero. Warn still fails CI via EX_DIAG_WARN. The tonal distinction is advisory: Warn means “this might be wrong”; Info means “FYI, probably intentional but worth surfacing”; Suggest means “I have a candidate replacement but I’m not confident enough to auto-apply it — eyes on it.” Rules like W034 sci-custom-control-info (which reports unpublished SCI control systems — legitimate per CAPCO but rare) are natural Info candidates; rules like S004 rel-to-trigraph-suggest (which proposes a higher-prior trigraph alternative for an ambiguous REL TO entry) emit at Suggest.

Suggest channel semantics

Suggest is the firing-but-non-applying channel: a diagnostic emitted at Suggest carries a candidate FixProposal that the engine will never auto-apply, regardless of confidence. The fix is informational — it tells the user what the rule would suggest if confidence were higher. Two paths produce Suggest-severity diagnostics:

1. Explicit emission: a rule constructs the diagnostic with Severity::Suggest directly. S004 rel-to-trigraph-suggest is the first such rule.
2. Engine rewrite: any diagnostic whose attached FixProposal has confidence.combined() < confidence_threshold is rewritten to Severity::Suggest by the engine in lint. This subsumes the prior silent-drop behavior at threshold-gate time so below-threshold proposals stay observable.

In both cases, Engine::fix filters out Suggest diagnostics from auto-apply by construction. Suggest diagnostics with fix: None are also valid (informational suggestion with no candidate replacement — used by future rules like #206’s REL TO opaque-uncertain reduction, where the rule has signal to surface but no specific replacement to propose); the renderer handles the missing-fix case cleanly.

Merge semantics (current: last-write-wins)

marque-config merges layers in strict precedence order — env vars override .marque.local.toml which overrides .marque.toml. Whatever the highest-precedence layer says for a given rule wins, including downgrades: a local override of "off" will suppress a project-config "error". This is intentional — individual classifiers sometimes need to silence a rule while iterating, and the audit log still records the configured severity for every applied fix.

If a future policy requires strictness-only merging (where a lower layer cannot downgrade a higher layer’s severity), change the loader to .max() over Severity::parse_config values rather than extend. The derived Ord above is already the correct operator for that case.

Variants

Off

Rule is disabled entirely. FR-008: severity=off is unrepresentable on emitted diagnostics — a rule at Off never fires, so no Diagnostic is produced.

Suggest

Advisory channel — diagnostic carries a candidate fix that will not auto-apply.

Distinct from Info (FYI, no actionable replacement) and from Off (non-firing). The fix-bearing diagnostic remains visible in lint output but the engine excludes it from auto-apply regardless of confidence. This is the suggest-don’t-fix channel: rules with low-confidence candidate corrections (e.g., S004 rel-to-trigraph-suggest) can surface “did you mean?” hints without committing to the rewrite.

Suggest keeps the CLI exit code at 0 (same as Info), so it is CI-silent.

Info

Emit informational diagnostic; does not block check-mode exit code. Intended for “audit-visible but probably intentional” signals — cases where the marking may be correct but the user may want to verify (e.g., unpublished SCI control systems).

Warn

Emit warning; non-error, but still non-zero in check mode (produces EX_DIAG_WARN = 2). Different from Info in tone and exit-code impact: Warn is “this might be wrong” and CI-visible; Info is “FYI, probably intentional but worth surfacing” and CI-silent (exit 0).

Error

Emit error; blocks --check exit code.

Fix

Apply fix automatically when --fix flag is present.

Implementations

impl Severity

pub fn parse_config(s: &str) -> Option<Self>

Parse a severity level from a config string. Returns None for unrecognized values; the config loader treats None as a hard error.

pub const fn as_str(self) -> &'static str

Canonical lowercase string form, suitable for JSON output.

This is the inverse of Severity::parse_config and is the stable surface that JSON consumers should depend on — never format!("{:?}") (which exposes Debug formatting as an unintended API).

Trait Implementations

impl Clone for Severity

fn clone(&self) -> Severity

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Severity

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

Formats the value using the given formatter.

impl Display for Severity

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

Formats the value using the given formatter.

impl Hash for Severity

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl Ord for Severity

fn cmp(&self, other: &Severity) -> Ordering

This method returns an Ordering between self and other.

fn max(self, other: Self) -> Self

where Self: Sized,

Compares and returns the maximum of two values.

fn min(self, other: Self) -> Self

where Self: Sized,

Compares and returns the minimum of two values.

fn clamp(self, min: Self, max: Self) -> Self

where Self: Sized,

Restrict a value to a certain interval.

impl PartialEq for Severity

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

Tests for self and other values to be equal, and is used by ==.

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

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl PartialOrd for Severity

fn partial_cmp(&self, other: &Severity) -> Option<Ordering>

This method returns an ordering between self and other values if one exists.

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

Tests less than (for self and other) and is used by the < operator.

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

Tests less than or equal to (for self and other) and is used by the <= operator.

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

Tests greater than (for self and other) and is used by the > operator.

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

Tests greater than or equal to (for self and other) and is used by the >= operator.

impl Copy for Severity

impl Eq for Severity

impl StructuralPartialEq for Severity