On 6/20/25 09:35, Dominique Devienne wrote:
> On Thu, Jun 19, 2025 at 6:35 PM Laurenz Albe <laurenz.albe@cybertec.at> wrote:
>>
>> On Thu, 2025-06-19 at 15:09 +0200, Dominique Devienne wrote:
>>> Hi. Little mystery we don't understand. v17.
>>> We're stumped for now.
>>
>> So are we. Why do you keep us guessing instead of posting a reproducer?
>
> Hi. Simply because there's too much proprietary stuff, I'm afraid.
> And it's likely some stupid mistakes on our part anyway. That I can't see...
> Still, the fact I see nothing extension-related in the libpq trace is
> intriguing, isn't it?
>
PQtrace logs client-server communication. I would not expect it to say
anything about actions that happen on the server, like for example
automatically dropping objects in a schema, after the schema is dropped.
I think the best way to move this forward is sharing a reproducer. If
you have too much proprietary stuff, you'll have to remove those bits,
or rather replace them with something you can share.
In fact, a reproducer is meant to be "minimal" - the smallest example
causing the issue. So creating reproducers generally means simplifying
the example as much as possible anyway. And I wouldn't be surprised if
in the process of doing that you find the answer yourself.
regards
--
Tomas Vondra
