Peter Eisentraut <peter_e@gmx.net> writes:
> Nope, the docs represent the behavior of the code at the time the docs
> were written. The code is now in error with respect to the documented
> behaviour. A quick check shows that PostgreSQL 7.0.2 agrees with
> including the fractional part.
[ checks ... ] As does 7.1. You're right, that is how it used to behave.
> For details, I refer you to my Oct 5 message "Unhappiness with forced
> precision conversion for timestamp", where we already discussed
> essentially the same issue, but apparently we never did anything about it.
I think the rest of us were waiting on Lockhart to opine about it ...
not to mention do something about it ...
regards, tom lane