Re: PL/pgSQL PERFORM with CTE
| От | David E. Wheeler |
|---|---|
| Тема | Re: PL/pgSQL PERFORM with CTE |
| Дата | |
| Msg-id | FC898158-D2F6-4DDC-A096-423FBC9BF4B2@justatheory.com обсуждение исходный текст |
| Ответ на | Re: PL/pgSQL PERFORM with CTE (Pavel Stehule <pavel.stehule@gmail.com>) |
| Список | pgsql-hackers |
On Aug 29, 2013, at 2:41 PM, Pavel Stehule <pavel.stehule@gmail.com> wrote: > I am thinking, so I propose a enough solution for you - when you use CTE for execution of VOID function, then result villbe VOID set, what we can accept as undefined result, and in this case a PERFORM should not be required. If CTE will returnsome result, then PERFORM should be required and PERFORM must to support CTE in all possible modes - updateable ornot updateable queries. If you can make PERFORM work with CTEs, that would be an improvement over the status quo. But I think there is no good reasonnot to let SELECT results be discarded, either. I know you think there are good reasons, but no one else in this threadis convince, AFAICT. Best, David
В списке pgsql-hackers по дате отправления: