Keary Suska <hierophant@pcisys.net> writes:
> Is it that PQgetResult will return a full query result from a single
> command, and will only return NULL if there are no more commands that have
> not completed, therefore only needing to be called once per command in the
> dispatched query; or is it that PQgetResult must be repeatedly called for
> even a single command query, and that it will only return NULL after each
> command in the dispatched query has completed fully?
You need to loop until PQgetResult returns null, if you want to handle
multi-command strings (you'll get one PGresult per command). If you
know there's only going to be one result you might be able to get away
without a second PQgetResult call, but I wouldn't recommend it.
> Further, can we *always* assume that query results will be returned in the
> order they appear in the query dispatch?
Yes. The backend isn't going to execute commands in a query string
out-of-order.
regards, tom lane