2010/11/9 Tom Lane <t...@sss.pgh.pa.us>: > Robert Haas <robertmh...@gmail.com> writes: >> On Sun, Nov 7, 2010 at 1:08 PM, Tom Lane <t...@sss.pgh.pa.us> wrote: >>> I guess I shoulda been paying closer attention :-(. That really, really >>> seems like fundamentally the wrong direction. What was it that was >>> unfixable about the other way? If it is unfixable, should we revert >>> ModifyTable? > >> The relevant thread is here: >> http://archives.postgresql.org/pgsql-hackers/2010-02/msg00783.php > > My opinion is still the same as here: > http://archives.postgresql.org/pgsql-hackers/2010-02/msg00688.php > > namely, that all we should be worrying about is a tuplestore full of > RETURNING tuples. Any other side-effects of a DML subquery should > *not* be visible to the calling query, and therefore all this argument > about snapshots and seqscan limits is beside the point.
Current consensus says: WITH x AS (SELECT count(*) FROM t), y AS (DELETE FROM t), z AS (SELECT count(*) FROM t) SELECT x.count, z.count FROM x, z; should return 0 for z.count but some number of original rows for x.count. So I think you need to read the underlying table itself as well as the emitted data of ModfyTable (or the result of any writeable CTE queries) in *predictable order*. To make it happen, we need CCI in each execution of child plans. Regards, -- Hitoshi Harada -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
