On 09/23/2014 06:23 PM, Andrew Dunstan wrote:
>
> On 09/23/2014 10:29 AM, Tom Lane wrote:
>> Robert Haas <robertmhaas@gmail.com> writes:
>>> Three people have voted for making it an *error* to supply a value
>>> that needs to be rounded, instead of changing the rounding behavior.
>> Votes or no votes, that's a horrible idea; it breaks the design goal
>> that users shouldn't need to remember the precise unit size when making
>> postgresql.conf entries.
>>
>> And I'm not sure what votes you're counting, anyway. People's opinions
>> have changed as the discussion proceeded ...
>
> I don't think I've weighed in on this, and yes, I'm very late to the
> party. The "round away from zero" suggestion seemed the simplest and
> most easily understood rule to me.
+1, rounding up seems better to me as well. Although throwing an error
wouldn't be that bad either.
- Heikki