Re: Oddity in EXPLAIN for foreign/custom join pushdown plans

I thought about the Relations line a bit more and noticed that there are cases where the table reference names for joining relations in the Relations line are printed incorrectly.  Here is an example:

postgres=# explain verbose select * from (select t1.a, t2.a from ft1 t1, ft2 t2 where t1.a = t2.a union select t1.a, t2.a from ft1 t1, ft2 t2 where t1.a = t2.a) as t(t1a, t2a);
                                                     QUERY PLAN
--------------------------------------------------------------------------------------------------------------------
 Unique  (cost=204.12..204.13 rows=2 width=8)
   Output: t1.a, t2.a
   ->  Sort  (cost=204.12..204.12 rows=2 width=8)
         Output: t1.a, t2.a
         Sort Key: t1.a, t2.a
         ->  Append  (cost=100.00..204.11 rows=2 width=8)
               ->  Foreign Scan  (cost=100.00..102.04 rows=1 width=8)
                     Output: t1.a, t2.a
                     Relations: (public.ft1 t1) INNER JOIN (public.ft2 t2)
                     Remote SQL: SELECT r1.a, r2.a FROM (public.t1 r1 INNER JOIN public.t2 r2 ON (((r1.a = r2.a))))
               ->  Foreign Scan  (cost=100.00..102.04 rows=1 width=8)
                     Output: t1_1.a, t2_1.a
                     Relations: (public.ft1 t1) INNER JOIN (public.ft2 t2)
                     Remote SQL: SELECT r1.a, r2.a FROM (public.t1 r1 INNER JOIN public.t2 r2 ON (((r1.a = r2.a))))
(14 rows)

The table reference names for ft1 and ft2 in the Relations line for the second Foreign Scan should be t1_1 and t2_1 respectively.

Would we really need the Relations line?  If joining relations are printed by core like "Foreign Join on public.ft1 t1_1, public.ft2 t2_1" as proposed upthread, we can see those relations from that, not the Relations line.

Also we can see the join tree structure from the deparsed query in the Remote SQL line.