Re: JDBC behaviour

On 18 February 2016 at 07:09, Mark Rotteveel <mark(at)lawinegevaar(dot)nl> wrote:

> On Thu, 18 Feb 2016 11:59:50 +0100 (CET), Andreas Joseph Krogh
> <andreas(at)visena(dot)com> wrote:
> > På torsdag 18. februar 2016 kl. 11:43:36, skrev Sridhar N Bamandlapally
> <
> > sridhar(dot)bn1(at)gmail(dot)com <mailto:sridhar(dot)bn1(at)gmail(dot)com>>:
> > The code/framework is written to handle batch inserts, which is common
> for
> > all
> > databases
> > I feel, PostgreSQL JDBC may need to modify setAutoCommit(false) code to
> > "implicit savepoint - on error - rollback to savepoint"
> >
> >
> > You simply cannot have batch-inserts in the same transaction and
> expecting
> > the
> > batch not to fail if one of the statements in the batch fails.
>
> On a lot of other database systems, that is exactly how it works. If a
> statement fails, that one statement is backed out (rolled back), and it is
> still up to the user to decide if he wants to commit or rollback the
> statements that did succeed.
>

This behaviour is an artifact of PostgreSQL. If you want to change the
transaction semantics of PostgreSQL then pgsql-hackers is the place to take
this up.

JDBC is just an interface. We aren't going to rewrite the backend semantics
to meet everyones needs/wants.

Dave Cramer

davec(at)postgresintl(dot)com
www.postgresintl.com