Re: [HACKERS] Almost there on column aliases

Thomas Lockhart <lockhart@alumni.caltech.edu> writes:
> Could someone run a "-d 99" query using the following from the
> regression test (rules.sql):
> select rtest_t2.a, rtest_t3.b
>  from rtest_t2, rtest_t3
>  where rtest_t2.a = rtest_t3.a;
> and send me the query, the rewritten query, and the plan emitted by
> the backend (it should be a MERGEJOIN plan)? It might speed up my
> rummaging around for the reason for the failure :(

This doesn't look very detailed, is it really what you wanted?

StartTransactionCommand
query: explain
select rtest_t2.a, rtest_t3.bfrom rtest_t2, rtest_t3where rtest_t2.a = rtest_t3.a
parser outputs:

{ QUERY :command 5  :utility ?  :resultRelation 0 :into <> :isPortal false :isBinary false :isTemp false :unionall
false:distinctClause <> :sortClause <> :rtable <> :targetlist <> :qual <> :groupClause <> :havingQual <> :hasAggs false
:hasSubLinksfalse :unionClause <> :intersectClause <> :limitOffset <> :limitCount <> :rowMark <>}
 

after rewriting:
{ QUERY   :command 5    :utility ?    :resultRelation 0   :into <>   :isPortal false   :isBinary false   :isTemp false
:unionall false   :distinctClause <>   :sortClause <>   :rtable <>   :targetlist <>   :qual <>   :groupClause <>
:havingQual<>   :hasAggs false   :hasSubLinks false   :unionClause <>   :intersectClause <>   :limitOffset <>
:limitCount<>   :rowMark <>  }
 

ProcessUtility: explain
select rtest_t2.a, rtest_t3.bfrom rtest_t2, rtest_t3where rtest_t2.a = rtest_t3.a
NOTICE:  QUERY PLAN:

Merge Join  (cost=164.66 rows=10000 width=12) ->  Sort  (cost=69.83 rows=1000 width=8)       ->  Seq Scan on rtest_t3
(cost=20.00rows=1000 width=8) ->  Sort  (cost=69.83 rows=1000 width=4)       ->  Seq Scan on rtest_t2  (cost=20.00
rows=1000width=4)
 

CommitTransactionCommand

> Another possibility is that I submit/commit my patches (there are
> quite a few files touched and I *really* want to get them off of my
> system and into the tree soon) but I was a bit hesitant to commit
> something with a known problem of this nature.

Any changes in backend/optimizer/ ?  I've got a bunch of uncommitted
changes there myself.
        regards, tom lane