On Tue, Apr 1, 2008 at 7:35 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Well, it's not like we have never before changed internal APIs in a
> minor update. (There have been security-related cases where we gave
> *zero* notice of such changes.) Nor am I willing to surrender the
> option to do so again. If there's somebody out there with a real
> problem with this change, they need to speak up.
+1 to fix it ASAP. A lot of people usually wait for 8.x.1 before
considering a migration and people are usually quite cautious with the
8.3 migration because of all the cast errors in the existing app.
Another question is how we can be sure it doesn't happen again. The
easiest way to test this is probably to have a JDBC test testing this
exact feature in the future benchfarm. Any comment?
--
Guillaume
