On Mon, 9 Oct 2023 at 21:00, Robert Haas <robertmhaas@gmail.com> wrote:
> ...but then...
>
> > With Citus the same user defined type can have
> > different OIDs on each of the servers in the cluster.
>
> I realize that your intention here may be to say that this is not an
> *additional* problem but one we have already. But it seems like one
> that we ought to be trying to solve, rather than propagating a
> problematic solution into more places.
Yes, I probably should have emphasized the word *additional*. i.e.
starting from scratch I wouldn't use OIDs in this GUC nor in
ParameterDescription or RowDescription, but blocking the addition of
this GUC on addressing that seems unnecessary. When we fix it we can
fix this too. I'd rather use OIDs (with all their problems)
consistently now for communication of types with regards to protocol
related things. Then we can at some point change all places in bulk to
some better identifier than OIDs.
> Decisions we make about the wire protocol are some of the most
> long-lasting and painful decisions we make, right up there with the
> on-disk format. Maybe worse, in some ways.
Yes, I agree. I just don't think using OIDs makes changing the
protocol in this regard any less painful than it already is currently.