Currently it's released when the close() method is called.
As for thread safe, it is. There's even a thread safety test in the
Maidstone Borough Council
Any views stated are my own, and not those of Maidstone Borough Council.
From: Rachel Greenham [mailto:rachel(at)enlarion(dot)demon(dot)co(dot)uk]
Sent: Thursday, June 24, 1999 2:13 PM
Subject: [INTERFACES] JDBC6.5-1.2 other possible problems
How/when does the JDBC actually release a postgres connection? I ask
because, when I try running stuff under native threads on Linux, unless
I explicitly close a connection it's left open after the Connection
object itself has gone out of scope - it's visible as a "postgres" task
in the task display, and isn't recovered until the JVM itself has been
killed. Also, it seems that if too many of them occur I get another
major error and the postgres back-end dies.
Platform (sorry, didn't mention in other post):
SuSE Linux 6.0 kernels 2.0.36 and 2.2.9, SMP and not, Blackdown Java
1.2, LiveSoftware's JRunPro servlet engine, Postgres 6.5. This problem
seems to manifest when running under native threads, which is not
*officially* working yet, but seems to.
Is the Postgres JDBC driver thread safe?
pgsql-interfaces by date
|Next:||From: Peter Mount||Date: 1999-06-24 14:41:41|
|Subject: RE: [INTERFACES] JDBC6.5-1.2 bug wrt timestamp|
|Previous:||From: Rachel Greenham||Date: 1999-06-24 13:12:58|
|Subject: JDBC6.5-1.2 other possible problems|