Re: New pg_lsn type doesn't have hash/btree opclasses

On 2014-05-10 19:08:48 -0400, Tom Lane wrote:
> Magnus Hagander <magnus(at)hagander(dot)net> writes:
> > On Sun, May 11, 2014 at 12:27 AM, Andres Freund <andres(at)2ndquadrant(dot)com>wrote:
> >> The problem is that once the beta is in progress (starting tomorrow),
> >> the projects tries to avoid changes which require a dump and restore (or
> >> pg_upgrade). Since the patch changes the catalog it'd require that.
>
> > It would be pg_upgrade'able though, wouldn't it? Don't we have precedents
> > for requiring pg_upgrade during beta? At least that's a smaller problem
> > than requiring a complete dump/reload.
>
> pg_upgrade makes the penalty for screwups smaller, but a post-beta1 initdb
> is still the result of a screwup. None of the historical examples you
> mention were planned in advance of beta.

Yea, I posted that just to answer Magnus' question.

I've argued that this omission should be fixed since tuesday. There's
been a tested and reviewed patch since
20140506230722(dot)GE24808(at)awork2(dot)anarazel(dot)de(dot) Given how many changes went
in since it certainly wouldn't have been a very destabilizing commit.

Anyway. I accept it's too late for beta1 now. Let's commit it if there's
another catversion bump.

Greetings,

Andres Freund

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