Bill Moseley wrote:
I'm not clear what the intended behavior is suppose to be. But it
seems that changing AutoCommit from 1 to 0 should issue a BEGIN, and
from 0 to 1 a COMMIT from the docs below.
Could someone clarify?
In testing changing AutoCommit does not issue and BEGIN or COMMIT.
IIRC issuing "begin work" to PostgreSQL may have undesirable
side-effects such as fixing the value of current_timestamp for the
duration of the transaction. So maybe it delays "begin work" until some
other statement is being executed?
It was quite a long time back that I encountered this... As you can
probably guess, I haven't tested it :D
SRH
