Re: EXPLAIN WITH

Stephen Frost <sfrost(at)snowman(dot)net> writes:
> * Stephen Frost (sfrost(at)snowman(dot)net) wrote:
>> Would be nice if there was a CTE ID or similar to link between
>> the pieces of the InitPlan and the CTE nodes.

> Erm, of course, the CTE *has* an ID already, since you name them. Could
> we get that ID/name up into the piece of the InitPlan that is handling
> that CTE?

I'm not sure but will be glad to take a look. Assuming it's not
unreasonably difficult, does anyone object to a format like this:

Nested Loop (cost=1266.59..1001458.19 rows=35936310 width=52)
InitPlan
CTE abc
-> Seq Scan on foo (cost=0.00..34.25 rows=647 width=12)
Filter: (id < 200)
CTE wumpus
-> Seq Scan on tab1 (cost=0.00..27.70 rows=1770 width=16)
-> CTE Scan on abc x (cost=0.00..35.40 rows=1770 width=16)
-> Materialize (cost=1204.64..1566.67 rows=20303 width=36)
-> Hash Join (cost=42.05..1025.34 rows=20303 width=36)
...

regards, tom lane