Tom Lane wrote:
> Ron Mayer <rm_pg@cheapcomplexdevices.com> writes:
>> Tom Lane wrote:
>>> We could try to do the same in the float case, but I'm a bit worried
>>> about finding ourselves showing "1234567.799999" ...
>> If I understand the code right [I didn't...]
>
> The problem is ... seconds field that includes hours,
> minutes, seconds, and fractional seconds...Here's an example...
> regression=# select '1234567890 hours 0.123 sec'::interval;
> ... 1234567890:00:00.123047
Hmm. That's also an existence proof that we're not too concerned
about showing 6 imprecise digits anyway (at least for some 8.3
DateStyles). Doesn't seem like it'd hurt too much if we show
them for all the IntervalStyles.
> Since there's a (somewhat arbitrary) limitation of the hours to 2^31,
> this is close to the worst possible case. (Hm, maybe someone actually
> did the math and decided that 2 fractional digits ...
Or I guess we could truncate to 2 digits only in the float case;
or truncate to 2 digits only if we're using the float case and
have large values. But that extra complexity doesn't seem
worth it to me - especially since it seems to only affect
people who do two non-default things (pick a date/interval style
that used to truncate to 2, and --disable-integer-datetimes).
I put a patch up at http://0ape.com/postgres_interval_patches
that does what I think seems getting reasonable. For better
or worse, it depends on the other two interval patches I was
working on, but I could make a version that doesn't depend on
those as well if people prefer that.