Re: [RFC] Removing "magic" oids

Greetings,

* Robert Haas (robertmhaas@gmail.com) wrote:
> On Sun, Oct 14, 2018 at 6:43 PM Andres Freund <andres@anarazel.de> wrote:
> > I'm not sure we want that however - yes, the short time pain will be
> > lower, but do we really want to inflict the confusion about invisible
> > oids on our users for the next 20 years? I think there's a fair argument
> > to be made that we should cause pain once, rather continuing to inflict
> > lower doeses of pain.
>
> Yeah, I think that argument has quite a bit of merit.  I suspect that
> there are a lot of people who expect that 'SELECT * FROM pg_whatever'
> is going to show them all of the data in pg_whatever, and take a while
> to figure out that it really doesn't.  I know better and still mess
> this up with some regularity.  Anyone who finds it unintuitive that *
> doesn't really mean everything is going to be happier with this change
> in the long run.
>
> In the short run, it is indeed possible that some catalog queries will
> break.  But, really, how many?  For the most part, automated queries
> written by tools probably select specific columns rather than
> everything, and IIUC those won't break.  And even if they do use
> 'SELECT *', won't they just end up with two copies of the OID column?
> That might work fine.
>
> Of course it might not, and then you'd have to fix your code, but it's
> not obvious to me that this would be a horror show.

I tend to agree that this won't be as much of a horror show as made out
to be up-thread.  Tools should certainly be using explicit column names
and if they aren't then we're breaking them regularly anyway whenever
we change what columns exist in a given catalog table.

All the column renaming we did for v10 strikes me as a much bigger deal
than this change and while we did hear some complaints about that, I
certainly feel like it was worth it and that people generally
understood.

Thanks!

Stephen