Re: pgsql: Have TRUNCATE update pgstat tuple counters

Tom Lane wrote:
> Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org> writes:
> > Have TRUNCATE update pgstat tuple counters
>
> > Catalog version bumped due to the change in pgstat format.
>
> I see no catversion.h change here, which is probably a good thing because
> no such bump should have been made for this change anyway. What *should*
> have been altered, and was not, was PGSTAT_FILE_FORMAT_ID. Please fix.

Right -- pushed that fix.

Now buildfarm member kouprey is failing the added test in
prepared_xacts. I am tempted to just remove it rather than try to fix
it: there's a 0.5s sleep in there (not enough for sure), but it would be
annoying to move it to the stats test which already sleeps however much
is necessary, because we would need to add an alternate output for the
case when prepared xacts are not enabled in the server.

--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services