Re: Is this the expected behaviour for DDL-query execution?

Thomas Johansson <thomas(dot)johansson(at)agama(dot)tv> writes:
> (detaild log message from pg_log
> 2009-05-15 00:00:17.179 CEST> LOCATION: make_inh_translation_lists,
> prepunion.c:992
> 2009-05-15 00:00:17.179 CEST> STATEMENT:
> UPDATE state_change SET (final_view_time, end_time) =
> (226, 10528) WHERE id = 91332641 AND time = 10523

> 2009-05-15 00:00:17.179 CEST> ERROR: XX000: could not find inherited
> attribute "id" of relation "state_change_20090430")

I'm still curious to see a complete test case for this behavior. AFAICS
the only way you could get that failure after an ALTER NO INHERIT would
be if the planner saw the changes to the child relation but did not see
the removal of the pg_inherits entry; which seems improbable.

regards, tom lane