>>> Kenneth Marshall <ktm@rice.edu> wrote:
>>> Even more surprising is the behavior for interval(1) here:
>>> [.... some context with nonsurprising examples removed ...]
>>> ccdev=# select '1 year 2 mons 3 days 04:05:06.64321'::interval(1);
>>> interval
>>> ----------------------------------
>>> 1 year 2 mons 3 days 04:05:06.60
>>> (1 row)
>>> That trailing zero should be considered a bug.
> What is not clear to me is how the
> decision to stop at the 2nd decimal digit was reached.
See this posting and others on the thread:
http://archives.postgresql.org/pgsql-hackers/2008-09/msg00999.php
The current rules seem to be:
(1) If precision is specified, round to that precision.
(2) If result has only zeros in the fraction, show no fraction, else
show at least two digits in the fraction, adding a trailing zero if
needed to get to two digits, but don't show any trailing zeros in the
fraction beyond the second position.
I think it would be ideal if we could track how many digits of
accuracy we have in a value, and show them all, even if that involves
trailing zeros. If that's not feasible, let's consistently not show
trailing zeros. Rounding .64 to .6 and then showing .60 is just plain
wrong.
-Kevin