On Thu, Nov 27, 2025 at 7:45 PM Aleksander Alekseev
<aleksander@tigerdata.com> wrote:
> > "Short byteas will have terminating NUL bytes in the abbreviated
> > datum. Abbreviated comparison need not make a distinction between
> > thse NUL bytes, and NUL bytes representing actual NULs in the
> > authoritative representation." [...]
> >
> > After that, the rest seems to flow better at a quick glance.
>
> Yes, it is much better now, thanks! Previously the comment was
> reasoning about NUL bytes as if normally bytea can't have them which
> IMO was confusing.
Pushed v7 with a few small adjustments, mostly for pgindent and the
new practice to prefer palloc_object().
--
John Naylor
Amazon Web Services