Skip to content
Sarj Code Standards

nan-hidden-by-builtin-min

Validate floating candidates before a built-in min/max can hide a NaN.

SARJ539 Rule source Tests Catalog JSON

Why

Python's min/max compares operands in order. A later NaN can be ignored, leaving a finite result that passes a finiteness check even though an original candidate was invalid.

Fix

Check every floating candidate before reducing it, or use a deliberately NaN-propagating reduction and validate its result.

Examples

Before — flagged Do not validate only the reduced value
app/export_validation.py
import math




def validate(single: float, batched: float) -> None:
    lowest = min(single, batched)
    if not math.isfinite(lowest) or lowest < 0.99:
        raise ValueError("Invalid export")
After — preferred Validate each floating candidate
app/export_validation.py
import math




def validate(single: float, batched: float) -> None:
    if not math.isfinite(single) or not math.isfinite(batched):
        raise ValueError("Invalid export")
    lowest = min(single, batched)
    if lowest < 0.99:
        raise ValueError("Invalid export")