Re: BLOB performance test FYI

> ...which IMHO implies multiple executions with the same parameters should
be
> possible

And is preferable. This is one of the benefits of using a prepared
statement -- doing loop constructs and such that don't force rebuilding it
all each time.

Clearly, JDBC has APIs built to support the concepts of small and large
binary objects. Most databases support them. I've been doing so with PG7.1
with success using OIDs, though I do need to run vacuumlo daily because my
JDBC code doesn't want to "know to unlink OIDs" since I'm trying to stay as
db agnostic as possible in my mainline code.

My impression was that blob support through JDBC in PG7.2 would be handled
by OID, and small binaries by bytea. Is that not the case? I know that I
have a "problem" in that I use setBytes/getBytes today for my blobs as that
worked in 7.1. I've not used the JDBC blob apis yet (setBlob()). I know
this changed in the PG JDBC code for 7.2, but it's still not clear to me how
this is all really working.

Thanks,
David