On Fri, 16 Jul 2010, Maciek Sakrejda wrote:
> Could it be that COPY is erroring out, and then the driver tries to
> end it anyway, causing the error? Anything interesting in the
> PostgreSQL logs?
There is nothing interesting in the Postgres logs at all to explain what
is going on. I have ERROR statements when I try to drop a table that isn't
there, but nothing when this COPY bugs out.
> Is it consistently reproducible with a specific data set?
>> For some reason, this seems to exhibit with some data but not with
>> others.
Yes, it seems to break with specific data. Particularly, it works with our
COPY tests, but it doesn't work with the higher-level tests.
So, the situation is that I have a database writing interface that I wrote
about six years ago, and we have been using a patched six-year-old
Postgres JDBC driver for ages. I have just updated to the newest JDBC
driver (which doesn't need to be patched for COPY), and I am now getting
this error. Could the binary format have changed slightly? (I haven't
changed my copy of Postgres when upgrading the JDBC driver - it is 8.4.4).
Thanks,
Matthew
--
There are only two kinds of programming languages: those people always
bitch about and those nobody uses. (Bjarne Stroustrup)