On 2021-Jun-09, Michael Paquier wrote:
> On Mon, Jun 07, 2021 at 12:44:42PM +0300, Aleksander Alekseev wrote:
> > I confirm that the bug still exists in master (be57f216). The patch
> > fixes it and looks good to me. I changed the wording a little and
> > added a regression test. The updated patch is in the attachment. I
> > added this thread to the CF and updated the status to "Ready for
> > Committer".
>
> FWIW, that looks rather natural to me to me to just ignore the object
> if it has already been dropped here. The same kind of rules apply to
> tables dropped with DROP TABLE which would not show up as of
> pg_event_trigger_ddl_commands(), but one can get a list as of
> pg_event_trigger_dropped_objects().
Oh, that parallel didn't occur to me. I agree it seems a useful
precedent.
> Alvaro, were you planning to look at that? I have not looked at the
> patch in details.
I have it on my list of things to look at, but it isn't priority. If
you to mess with it, please be my guest.
> missing_ok is available in getObjectIdentity() only
> since v14, so this cannot be backpatched :/
Ooh, yeah, I forgot about that. And that one was pretty invasive ...
I'm not sure if we can reasonably implement a fix for older releases.
I mean, it's a relatively easy test: do a syscache search for the object
or a catalog indexscan (easy to do with get_object_property_data-based
API), and if the object is gone, skip getObjectTypeDescription and
getObjectIdentity. But maybe this is too much code to add to stable
releases ...
--
Álvaro Herrera Valdivia, Chile