Re: Synthesize support for Statement.getGeneratedKeys()?
От | Michael Paesold |
---|---|
Тема | Re: Synthesize support for Statement.getGeneratedKeys()? |
Дата | |
Msg-id | 45810230.80803@gmx.at обсуждение исходный текст |
Ответ на | Synthesize support for Statement.getGeneratedKeys()? (Ken Johanson <pg-user@kensystem.com>) |
Ответы |
Re: Synthesize support for Statement.getGeneratedKeys()?
Re: Synthesize support for Statement.getGeneratedKeys()? |
Список | pgsql-jdbc |
Ken Johanson wrote: > Hello, > > I've just come to realize that Statement.getGeneratedKeys() is not > supported in the current PG and/or JDBC driver. Does someone know if > this is a limitation of PG, or its protocol, or just not yet implemented > in the JDBC driver? I'm just wondering where, if at all (if I have > enough brain cells that is :), I could try to offer a patch or ideas.. It would be possible to implement that using the RETURNING clause supported in recent versions of the server. See here: http://archives.postgresql.org/pgsql-jdbc/2006-10/msg00035.php http://archives.postgresql.org/pgsql-jdbc/2006-10/msg00037.php Unfortunately, all my available spare time went into implementing support for standards_conforming_strings in the 8.2 release cycle. > Then question 2; I did see a discussion where it was suggested that we > could get roughly the same effect by issuing a > SELECT currval('<sequence-name>'); after the DML... > http://archives.postgresql.org/pgsql-jdbc/2005-10/msg00035.php > Would it then be feasible internal to the JDBC driver to (create an > option that would enable) always implicitly append that query to the > (String)sql arg of Statment.executeUpdate(String sql, String[] > columnNames)? I mean, just internally attempt to create the same > behavior as what this method should be doing? As was discussed before, a solution just for sequences is not general enough, therefore the RETURNING thing. > Question 3 is, it seems like option two would only return the last > created key, not set of keys, in the case where multiple rows were > inserted.. is this accurate? Seems correct. And that's one of the objections to the idea. > Unfortunately if I cant find a way to make my target-app work with PG > (without adding PG-specific modifications for getting keys), I'm > probably not going to be able to make the switch to PG unfortunately - > the code I'm working with makes really, really extensive use of > retrieved keys.. Could you comment on my mail above (the second one) -- what API-methods does your software use, actually? Best Regards Michael Paesold
В списке pgsql-jdbc по дате отправления: