Peter Eisentraut <peter.eisentraut@2ndquadrant.com> writes:
> On 11/02/2019 21:33, Tom Lane wrote:
>> I don't have time to probe further right now, but I believe what is
>> wrong is that we're missing a CommandCounterIncrement call after the
>> sequence is created, or at least after its internal dependency on the
>> table column is created. When the transaction commits immediately,
>> dependency.c fails to see the internal dependency entry and so it
>> doesn't remove the sequence. Wrapped in a transaction, there's a
>> CCI somewhere and all is well.
> Right. I think it would be good to put a CommandCounterIncrement() at
> the top of PreCommit_on_commit_actions(). That ensures that the
> dependency code always see the latest state.
Hm, I'd be more inclined to find where the sequence creation is happening
and add a CCI at the end, because that comports better with the general
plan for inserting CCIs. There may be other issues of this same sort with
doing-X-just-after-identity-sequence-creation if you don't fix it that
way.
>> That bollixes later attempts to clean out the temp
>> namespace, since deletion tries to recurse to the missing table.
> Should there be some warnings when this happens?
Like what?
regards, tom lane