diff options
| author | Peter Eisentraut <peter@eisentraut.org> | 2022-04-19 20:38:53 +0200 |
|---|---|---|
| committer | Peter Eisentraut <peter@eisentraut.org> | 2022-04-19 21:04:52 +0200 |
| commit | f2a2bf66c87e14f07aefe23cbbe2f2d9edcd9734 (patch) | |
| tree | 8ee986db7380f57b38eee0d5d368caa0d2662d6e /src/backend/utils/adt/timestamp.c | |
| parent | a62bff74b135b191de0417d1cc8ec52f144c8fe7 (diff) | |
| download | postgresql-f2a2bf66c87e14f07aefe23cbbe2f2d9edcd9734.tar.gz postgresql-f2a2bf66c87e14f07aefe23cbbe2f2d9edcd9734.zip | |
Fix extract epoch from interval calculation
The new numeric code for extract epoch from interval accidentally
truncated the DAYS_PER_YEAR value to an integer, leading to results
that mismatched the floating-point interval_part calculations.
The commit a2da77cdb4661826482ebf2ddba1f953bc74afe4 that introduced
this actually contains the regression test change that this reverts.
I suppose this was missed at the time.
Reported-by: Joseph Koshakow <koshy44@gmail.com>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://www.postgresql.org/message-id/flat/CAAvxfHd5n%3D13NYA2q_tUq%3D3%3DSuWU-CufmTf-Ozj%3DfrEgt7pXwQ%40mail.gmail.com
Diffstat (limited to 'src/backend/utils/adt/timestamp.c')
| -rw-r--r-- | src/backend/utils/adt/timestamp.c | 14 |
1 files changed, 10 insertions, 4 deletions
diff --git a/src/backend/utils/adt/timestamp.c b/src/backend/utils/adt/timestamp.c index da73796eac8..552b631ba78 100644 --- a/src/backend/utils/adt/timestamp.c +++ b/src/backend/utils/adt/timestamp.c @@ -5308,10 +5308,16 @@ interval_part_common(PG_FUNCTION_ARGS, bool retnumeric) int64 secs_from_day_month; int64 val; - /* this always fits into int64 */ - secs_from_day_month = ((int64) DAYS_PER_YEAR * (interval->month / MONTHS_PER_YEAR) + - (int64) DAYS_PER_MONTH * (interval->month % MONTHS_PER_YEAR) + - interval->day) * SECS_PER_DAY; + /* + * To do this calculation in integer arithmetic even though + * DAYS_PER_YEAR is fractional, multiply everything by 4 and then + * divide by 4 again at the end. This relies on DAYS_PER_YEAR + * being a multiple of 0.25 and on SECS_PER_DAY being a multiple + * of 4. + */ + secs_from_day_month = ((int64) (4 * DAYS_PER_YEAR) * (interval->month / MONTHS_PER_YEAR) + + (int64) (4 * DAYS_PER_MONTH) * (interval->month % MONTHS_PER_YEAR) + + (int64) 4 * interval->day) * (SECS_PER_DAY / 4); /*--- * result = secs_from_day_month + interval->time / 1'000'000 |