The patch I sent earlier is flaud with respect to subplan parameter
numbering, I counted from zero where the parParam list had to be used.
Yeb Havinga wrote:
> See patch below against HEAD.
>
> Example of query against catalog:
>
> postgres=# explain verbose select oid::int + 1,(select oid from
> pg_class a where a.oid = b.relfilenode and a.relnamespace =
> b.relnamespace) from pg_class b;
> QUERY
> PLAN
> --------------------------------------------------------------------------------------------------------
>
> Seq Scan on pg_catalog.pg_class b (cost=0.00..2459.64 rows=296 width=12)
> Output: ((b.oid)::integer + 1), SubPlan 1 ($0 := b.relfilenode, $1
> := b.relnamespace)
> SubPlan 1
> -> Index Scan using pg_class_oid_index on pg_catalog.pg_class a
> (cost=0.00..8.27 rows=1 width=4)
> Output: a.oid
> Index Cond: (a.oid = $0)
> Filter: (a.relnamespace = $1)
> (7 rows)
>
> Explain of alternative subplan (query comes from aggregates.sql in
> regression).
>
> regression=# explain verbose select ten, sum(distinct four) from onek a
> group by ten
> having exists (select 1 from onek b where sum(distinct a.four) = b.four);
> QUERY
> PLAN
> --------------------------------------------------------------------------------------
>
> GroupAggregate (cost=197.83..216.52 rows=10 width=8)
> Output: a.ten, sum(DISTINCT a.four)
> Filter: (alternatives: SubPlan 1 ($0 := sum(DISTINCT a.four)) or
> hashed SubPlan 2)
> -> Sort (cost=197.83..200.33 rows=1000 width=8)
> Output: a.ten, a.four
> Sort Key: a.ten
> -> Seq Scan on public.onek a (cost=0.00..148.00 rows=1000
> width=8)
> Output: a.ten, a.four
> SubPlan 1
> -> Seq Scan on public.onek b (cost=0.00..150.50 rows=250 width=0)
> Filter: ($0 = b.four)
> SubPlan 2
> -> Seq Scan on public.onek b (cost=0.00..148.00 rows=1000 width=4)
> Output: b.four
> (14 rows)
>
> Would the explain above be better if the filter with subplans
> arguments showed the EXISTS keyword? Much code from get_sublink_expr
> could be reused to show additional info depending on the sublink type
> and testexpr.
>
> Other info: there are now only appends to the context->buf. The
> problems with OUTER var I mentioned earlier are gone now arguments are
> deparsed using get_rule_expr instead of deparse_expression.
>
> regards,
> Yeb Havinga
>
