"Aleksandrs Vinogradovs" <avinogradovs@xxxxxxxxxxxxxxxx> writes: > How do I handle SPI errors properly without rethrowing them or > using subtransactions ? Those are your only two choices. > My current code works (simple PG_CATCH() > with resseting of error state), but reports reference leaks for > every failed SPI_exec() call... The warnings are the least of your worries. Without a subxact abort, there is all sorts of state that has not been properly cleaned up. regards, tom lane ---------------------------(end of broadcast)--------------------------- TIP 2: Don't 'kill -9' the postmaster
