#Allow

#Allow("rule") switches off one rux lint rule for the declaration below it, and nowhere else. It is for a declaration with a deliberate reason to break a convention — most often a name that must match an outside contract, such as a C struct or a C macro.

allow-attribute = "#Allow" "(" string-literal ")"

Rules

// Mirrors the C struct, so it keeps C's spelling.
#Allow("naming.type")
struct timespec {
    seconds: int64;
    nanoseconds: int64;
}

// A C macro's own name, which a reader porting code will look for.
#Allow("naming.const")
const SEEK_SET: int32 = 0;

Without the attributes, rux lint reports:

warning: struct name 'timespec' should be PascalCase
  help: rename it to 'Timespec'
warning: constant name 'SEEK_SET' should be PascalCase
  help: rename it to 'SeekSet'

One #Allow names one rule; stack several for several rules. The rule names are checked when the source is compiled, not only when it is linted, so a typo cannot silence nothing:

See also

• rux lint — the checks these rules belong to
• Identifiers — the naming conventions
• Learn: Compile error, Tooling