Re: PANIC in heap_delete during ALTER TABLE

On Thu, Sep 22, 2022 at 11:47 AM Jeff Davis <pgsql(at)j-davis(dot)com> wrote:
> Committed.

My thoughts on the fix (a little belated):

The patch looks very much like it can only affect cases that are going
to crash anyway (by forcing a restart). Plus you're just adding
another reason to restart in heap_delete() -- there were already a few
reasons why that might be necessary. So the fix is very well targeted.
Even when I think about it in the most adversarial/pessimistic way
possible, it's still very difficult to imagine anybody being worse off
than they would have been without the patch. OTOH it's readily
apparent that this fixes a crash.

There are no absolute certainties, but I do feel confident that
committing this fix was the right response, all things considered.

--
Peter Geoghegan