Kris Jurka wrote:
> The first time through it does not fail because the driver
> needs to query the backend to get some setup information for large objects
> which starts a transaction.
Sounds like that is (another) bug .. it should be using the
QUERY_SUPPRESS_BEGIN flag for driver-generated queries to avoid starting
a transaction accidentally. I fixed that in various other places but
didn't think to check the LO code.
Also, any thoughts on making the LO vs. bytea behaviour a separate
option, rather than lumping it in with 7.1 compatibility? It seems quite
possible that you might want to use LOs for get/setBytes() but use the
most up to date driver behaviour elsewhere.
-O