Re: [HACKERS] scan on inheritance parent with no children in current session

On Sun, Aug 6, 2017 at 10:56 PM, Amit Langote
<Langote_Amit_f8@lab.ntt.co.jp> wrote:
> Good catch.  I agree that getting an Append node after that commit is
> unintentional and we should fix so that we don't get an Append.  So, +1 to
> your patch.  I looked at the patch and the code fix seems to do what we want.

So, I also agree that this is a good fix, but I don't think it fixes
the whole problem.  Consider:

rhaas=# create table parent (a int) partition by list (a);
CREATE TABLE
rhaas=# create temp table child partition of parent for values in (1);
CREATE TABLE
rhaas=# explain verbose select * from parent;                              QUERY PLAN
-------------------------------------------------------------------------Append  (cost=0.00..35.50 rows=2550 width=4)
-> Seq Scan on pg_temp_3.child  (cost=0.00..35.50 rows=2550 width=4)        Output: child.a
 
(3 rows)

But the comments say:
* A childless table is never considered to be an inheritance set; therefore* a parent RTE must always have at least two
associatedAppendRelInfos.
 

Yet, not.  So at least the comments need to be updated; not sure if we
want to try to eliminate the Append node in this case also.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company