Thread: pgbouncer transaction patch?

Happened upon this:

http://treehou.se/~omar/postgresql-jdbc-8.4-701-pgbouncer_txn.patch

and the related (ancient) pgbouncer mailing list post

http://pgfoundry.org/pipermail/pgbouncer-general/2010-February/000507.html

It seems sensible, but as far as I can tell, no attempt had ever been
made to upstream this. Any thoughts on accepting this? The pgbouncer
FAQ still references this issue:

http://pgbouncer.projects.postgresql.org/doc/faq.html

I pushed it out to my github repo for easier merging (to be clear: I
did not author the patch; I merely found something that, at first
blush, seems to belong in the driver; I thought I'd take a stab at
getting it in):

https://github.com/deafbybeheading/pgjdbc/tree/pgbouncer-txn

The original author appears to be the same person who sent the mailing
list post.

For what it's worth, I did run the tests (on Java 6); everything
passes. I don't think the change warrants any new tests.

On 08/15/2012 12:27 PM, Maciek Sakrejda wrote:
> Happened upon this:
>
> http://treehou.se/~omar/postgresql-jdbc-8.4-701-pgbouncer_txn.patch

Is there any reason PgJDBC should ever NOT set
QueryExecutor.QUERY_ONESHOT for COMMIT / ROLLBACK / BEGIN ? I wonder if
this should just be made unconditional.

--
Craig Ringer