Use NUMERIC for money, never float

Use NUMERIC for money, never float | BeforeMerge Rules | BeforeMerge

Why this matters

real/double precision are binary floating point and cannot store values like 0.10 exactly, so monetary sums accumulate rounding error and equality checks misfire. numeric (a.k.a. decimal) is exact base-10 with defined precision and scale — the correct type for money.

-- bad
amount double precision
 
-- good
amount numeric(12,2) NOT NULL

Why This Matters

Related Rules

Catch this automatically on every PR

Why this matters