Roman Kononov <kononov195-pgsql(at)yahoo(dot)com> writes:> In float4mul() and float4div(), the computation should be double precision.
Why? It's going to have to fit in a float4 eventually anyway.
regards, tom lane