Re: Reducing runtime of stats regression test

Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> On Thu, May 4, 2017 at 10:22 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> Yes, but that would be getting into the realm of new features, not
>> post-feature-freeze test adjustments. It certainly couldn't be
>> a candidate for back-patching.

> I'm not sure there's some bright line between adding a new
> SQL-callable function to cut down the test time and any other
> tinkering we might do to reduce the regression test time. I think
> there's a pretty good argument that all of the recent changes you made
> in this area constitute strictly optional tinkering. I'm haven't been
> objecting because they don't seem likely to destabilize anything, but
> I don't see that they're really helping us get ready for beta either,
> which is presumably what we ought to be focusing on at this point.

Well, to my mind, making the regression tests faster is something that
could be quite helpful during beta, because a lot of people will be
running them. (Or so we hope, at least.) If the fact that e.g.
the recovery tests take a lot of time discourages people from running
them, that can't be a good thing for beta purposes. So I respectfully
reject your opinion about what I should be working on.

regards, tom lane