Bruce Momjian <maillist@candle.pha.pa.us> writes:
> Bug confirmed. Wow, that is strange. There isn't anything about temp
> table that would suggest this would happen.
I see it too. explain shows something pretty fishy:
regression=> explain update tmpArray set val[3] = 7;
NOTICE: QUERY PLAN:
Nested Loop (cost=43043.00 rows=1000000 width=22) -> Seq Scan on pg_temp.2904.0 (cost=43.00 rows=1000 width=12) ->
SeqScan on tmparray (cost=43.00 rows=1000 width=10)
EXPLAIN
I'm betting that something in the array code is somehow bypassing the
normal table lookup mechanism, and is managing to see the underlying
temp-table name that should be hidden from it. Will look further...
regards, tom lane