Thread
-
'epoch'::timestamp and Daylight Savings
Hosen, John <john.hosen@capita.co.uk> — 2002-10-22T16:11:35Z
Hi there all PostgreSQL Hackers. We have just upgraded from 7.1.3 to 7.2.3 over the weekend, and have just noticed something weird with regards 'epoch'::timestamp. In 7.1.3, 'epoch'::timestamp always returned the value 1970-01-01 00:00:00+00, following the upgrade we now get:- timestamptz ------------------------ 1970-01-01 01:00:00+01 Also, any fields set to 'epoch'::timestamp before the upgrade now return:- e_app_xfer_date ------------------------ 1970-01-01 00:00:00+01 If we issue a SET TIMEZONE TO 'GMT' the results are the same as for the previous release. This is causing us a problem as we use epoch to indicate whether the data has been processed, and to log the date and time of processing. We are running on RedHat 7.1 with kernel 2.4.9-21smp, and glibc glibc-2.2.4-19.3 with PostgreSQL compiled from the source tarfile. Output from version(): PostgreSQL 7.2.3 on i686-pc-linux-gnu, compiled by GCC 2.96 With regards PostgreSQL in general, we have been using it since version 6.4.x and have found it (even the earier versions) extremely reliable an powerful. Keep up the good work guys! John Hosen Senior Network Analyst Capita RAS http://www.capitaras.co.uk/ *********************************************************************** This email and any files attached to it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify the system manager. The message and any files attached to it have been scanned by MIMEsweeper with Sophos Sweep and found to be free from all known viruses. Information on MIMEsweeper can be found at http://www.mimesweeper.com/ *********************************************************************** -
Re: 'epoch'::timestamp and Daylight Savings
lockhart@fourpalms.org — 2002-10-23T15:06:09Z
> We have just upgraded from 7.1.3 to 7.2.3 over the weekend, and have just > noticed something weird with regards 'epoch'::timestamp. > In 7.1.3, 'epoch'::timestamp always returned the value 1970-01-01 > 00:00:00+00,... I would expect 'epoch'::timestamp to always act as though the value were the same as Unix time zero. But it did not explicitly return that value: lockhart=# select version(); ------------------------------------------------------------- PostgreSQL 7.1.2 on i686-pc-linux-gnu, compiled by GCC 2.96 lockhart=# select 'epoch'::timestamp; ---------- epoch lockhart=# select 'epoch'::timestamp + '0 sec'::interval; ------------------------ 1969-12-31 16:00:00-08 > following the upgrade we now get:- > timestamptz > ------------------------ > 1970-01-01 01:00:00+01 Which is consistant with the previous result in absolute time. You will find that the most recent versions of PostgreSQL convert 'epoch' on input, and that with- and without time zone data types are available: lockhart=# select cast('epoch' as timestamp with time zone); timestamptz ------------------------ 1969-12-31 16:00:00-08 lockhart=# select cast('epoch' as timestamp without time zone); timestamp --------------------- 1970-01-01 00:00:00 > Also, any fields set to 'epoch'::timestamp before the upgrade now return:- > e_app_xfer_date > ------------------------ > 1970-01-01 00:00:00+01 > If we issue a SET TIMEZONE TO 'GMT' the results are the same as for the > previous release. Not sure about this one. What is the schema? Can you give an example where the time gets shifted by an hour? > This is causing us a problem as we use epoch to indicate whether the data > has been processed, and to log the date and time of processing. Using "special values" to indicate status can be troublesome, as you are finding. I'd suggest using NULL to indicate that a field is not known or not yet set. > This email and any files attached to it are confidential and intended > solely for the use of the individual or entity to whom they are > addressed. If you have received this email in error please notify > the system manager. Oops. Can you please confirm that we were the intended recipients? ;) hth - Thomas