Re: let's make the list of reportable GUCs configurable (was Re: Add %r substitution for psql prompts to show recovery status)

On 11 January 2018 at 09:55, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:

> Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> > But if we add this feature and somebody wants to use it for
> > server_version_num, it's really pretty simple. In the startup packet,
> > you say _pq_.report=server_version_num. Then, you call
> > PQparameterStatus(conn, "server_version_num"). If you don't get a
> > value, you try calling PQparameterStatus(conn, "server_version") and
> > extracting the second word. If that still doesn't work then you give
> > up. That doesn't seem either useless or difficult to implement
> > correctly from here.
>
> Yeah, but what's the point? If yuou have to maintain the server_version
> parsing code anyway, you're not saving any complexity with this. You're
> just creating a code backwater that you probably won't test adequately.
>
>
If we'd done server_version_num in 9.5, for example, less stuff would've
broken with pg10.

I just don't get it. The argument you use can be applied to almost any
change, to say "why bother making change <x> because people can't rely on
it for <y> years, so it's pointless to have it at all". Why add protocol v3?

PostgreSQL is a stable, long term project, and I'd like to plan for the
future. I also think people are way more likely to handle things like
--with-extra-version correctly when dealing with server_version_num, where
I don't much *care* if code parsing old server versions breaks.

--
Craig Ringer http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services