Vaclav TVRDIK <tvrdik@i3.cz> writes:
> I have one question about converting timestamps to text. I believed
> that using to_char function with proper mask and casting by style
> COLUMN::text is equal, but when I issue following query against 8.3.3
> database it returns me different values (they differ on last position
> of microsecconds value, maybe rounding error ?).
Hm, I can't replicate that here --- but if you're using floating-point
timestamps, as is default in 8.3, then a certain amount of
machine-dependent roundoff fuzziness is not surprising.
regards, tom lane