Re: Problem with datetime in updates
| Newsgroups | gmane.comp.gnu.enterprise.general |
|---|---|
| Message-ID | <[email protected]> |
El lun, mar 29, a las 01:08:55 James Thompson escribía: > > > > > id | nick | name | created | ... > > ----+------+------+----------------------------+ ... > > 11 | f | ff | 2004-03-29 01:23:22.886056 | ... > > ^^^^^^ > > status='prop' > > > > i.e.: no decimal digits for seconds, so none record gets updated. > > > > ¿Any suggestion about how can I deal with it? > > Yes. I've seen this as well and forgot this would impact SuSE installs. It > seems that the pygresql returns timestamps as rounded strings. Since > gnue-common cannot assume that the underlying database provides any type of > unique row id or a primary key it performs the update using the original > values retrieved in the where clause to find the correct row. The rounded > string causes that match to fail on any table w/ a timestamp field. > Newbie issue: Definition of table had primary key declared, but datasource in form not. Including datasource/@primarykey solved the problem. As a future suggestion, I've found that with postgresql a subscript for timestamp type that says the number of decimal digits may be used. If schema definition could include an atribute for that, tables without primary key would benefit. I think it's not good for db drivers rounding decimal digits, i.e., alter stored values, without programmer 'permission'. thanks for help. -- José Esteban Granada - Spain