Re: Timestamp with libpq
От | Jakob Lechner |
---|---|
Тема | Re: Timestamp with libpq |
Дата | |
Msg-id | 1223901242.3502.54.camel@vie063.fabagl.fabasoft.com обсуждение исходный текст |
Ответ на | Re: Timestamp with libpq (Michael Meskes <meskes@postgresql.org>) |
Ответы |
Re: Timestamp with libpq
|
Список | pgsql-interfaces |
Hi Michael, Am Montag, den 13.10.2008, 12:46 +0200 schrieb Michael Meskes: > Please keep in mind that there is no guarantee that your server sends a double > in a binary query. This depens on whether integer-datatypes are configure or > not. Or in other words, it might be a long long instead of a double. Yes, you're right. The SLES postgres server transmits timestamps as long long numbers (microseconds since 2000-01-01). > Is there any reason to use a binary transfer? If I use textual transfer I'm losing precision. E.g. for timestamps the returned string from my table is "1955-06-08 00:00:00". Thus I'm restricted to timestamps with a granularity of 1 second. > I would not recommend this in a > general setup. If you just want to avoid some ascii translation hassle, how > about using ecpg instead of libpq? The generic database interface I'm writing is part of a basic library used in our project. The interface serves as an abstraction for accessing different database servers (Postgres, MSSQL, ...). I'm currently working on the postgres implementation of the interface. As far as I've seen using ecpg means to hardcode SQL statements which of course can't be done in a generic library. Thanks for your helpful hints. But it's seems I don't have much choice but to use ascii transfer mode. Best regards Jakob -- Jakob Lechner Research & Development appl.strudl Software GmbH Honauerstraße 4 A-4020 Linz Tel.: [+43] (70) 60 61 62 Fax: [+43] (70) 60 61 62-609 E-Mail: jakob.lechner@applstrudl.com Web: http://www.applstrudl.com Handelsgericht Linz, FN 303988 t
В списке pgsql-interfaces по дате отправления: