On Wed, Sep 21, 2011 at 12:19 PM, Andres Freund <andres@anarazel.de> wrote:
> /*
> * We also disallow data-modifying WITH in a cursor. (This could be
> * allowed, but the semantics of when the updates occur might be
> * surprising.)
> */
> if (result->hasModifyingCTE)
> ereport(ERROR,
> (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
> errmsg("DECLARE CURSOR must not contain data-modifying
> statements in WITH")));
>
> Given that cursors are about the only sensible way to return larger amounts of
> data, that behaviour reduces the usefulness of wCTEs a bit.
>
> Whats the exact cause of concern here? I personally don't think there is a
> problem documenting that you should fetch the cursor fully before relying on
> the updated tables to be in a sensible state. But that may be just me.
Well, it looks like right now you can't even using a simple INSERT ..
RETURNING there:
rhaas=# create table wuzzle (a int);
CREATE TABLE
rhaas=# declare w cursor for insert into wuzzle select g from
generate_series(1, 10) g returning g;
ERROR: syntax error at or near "insert"
LINE 1: declare w cursor for insert into wuzzle select g from genera... ^
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company