Re: Odd behaviour of timestamptz
От | hubert depesz lubaczewski |
---|---|
Тема | Re: Odd behaviour of timestamptz |
Дата | |
Msg-id | 9e4684ce0702210150m35d5626fx9f3d27c1726edc03@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: Odd behaviour of timestamptz (Michael Glaesemann <grzm@seespotcode.net>) |
Список | pgsql-general |
On 2/21/07, Michael Glaesemann <grzm@seespotcode.net> wrote:
doesn't seem to be relevant.
i got brand new cvs head (about 40 minutes ago). configured it with:
./configure \
--prefix=/home/pgdba/work \
--with-pgport=5810 \
--with-tcl \
--with-perl \
--with-python \
--enable-integer-datetimes \
--without-krb5 \
--without-pam \
--without-bonjour \
--with-openssl \
--with-readline \
--with-zlib \
--with-gnu-ld
and rerun the test:
# create TABLE test (data timestamptz);
CREATE TABLE
# INSERT into test values ('1910-01-10');
INSERT 0 1
# INSERT into test values ('1990-01-10');
INSERT 0 1
# select * from test;
data
---------------------------
1910-01-10 00:00:00+01:24
1990-01-10 00:00:00+01
(2 rows)
still something's wrong.
depesz
I suspect your RPMs build PostgreSQL without --enable-integer-
datetimes. Without this configure flag, timestamps are represented as
floats, with all of the imprecision that implies. See the second note
below the Date/Time Types table:
doesn't seem to be relevant.
i got brand new cvs head (about 40 minutes ago). configured it with:
./configure \
--prefix=/home/pgdba/work \
--with-pgport=5810 \
--with-tcl \
--with-perl \
--with-python \
--enable-integer-datetimes \
--without-krb5 \
--without-pam \
--without-bonjour \
--with-openssl \
--with-readline \
--with-zlib \
--with-gnu-ld
and rerun the test:
# create TABLE test (data timestamptz);
CREATE TABLE
# INSERT into test values ('1910-01-10');
INSERT 0 1
# INSERT into test values ('1990-01-10');
INSERT 0 1
# select * from test;
data
---------------------------
1910-01-10 00:00:00+01:24
1990-01-10 00:00:00+01
(2 rows)
still something's wrong.
depesz
В списке pgsql-general по дате отправления: