We have one query that uses postgis and it returns bytea stuff, so we'll need to set
bytea_output=escape
as default in our configuration file until the drivers are updated. But sounds like that's the main issue for us.
Hopefully this will go easy. I haven't run into any problems yet, I'm now going through the tuning we did for 8.3 and verifying what tuning makes sense still, but I'm happy to see only a few small bumps in regards to the JDBC driver.
Thanks
On Thu, Feb 16, 2012 at 2:06 PM, Kris Jurka
<books@ejurka.com> wrote:
On Thu, 16 Feb 2012, Ned Wolpert wrote:
> So, the driver issue was only in dealing with UNICODE issues? Is it true
> that if we are always set with UNICODE client_encoding we should be fine?
> (Or, another way to put it, we have already tested connection from the
> existing java apps using the 8.3 driver to the new 9.1 database test
> instance, and they worked 'just fine' for initial queries... is that where
> this problem would manifest itself? Only in connecting?) Or was there
> something else I missed in that thread?
The end result of that thread was that the server folks ended up fixing
the server to allow older JDBC clients to connect, so that was only an
issue in the development version of the server for a short period of
time, never in a released version.
The big backwards compatibility change that occurred between 8.3 and 9.1
servers was the bytea escaping option. If you want to use an older driver
you must set the bytea_output config option to escape on the server.
http://www.postgresql.org/docs/9.1/static/runtime-config-client.html#GUC-BYTEA-OUTPUT
Kris Jurka
--
Virtually, Ned Wolpert
http://codehead.tumblr.com/http://codeheadsystems.wordpress.com/"Settle thy studies, Faustus, and begin..." --Marlowe