Re: Some improvements to numeric sqrt() and ln()

Dean Rasheed <dean(dot)a(dot)rasheed(at)gmail(dot)com> writes:
> On Wed, 4 Mar 2020 at 14:41, David Steele <david(at)pgmasters(dot)net> wrote:
>> Are these improvements targeted at PG13 or PG14? This seems a pretty
>> big change for the last CF of PG13.

> Well of course that's not entirely up to me, but I was hoping to
> commit it for PG13.

> It's very well covered by a large number of regression tests in both
> numeric.sql and numeric_big.sql, since nearly anything that calls
> ln(), log() or pow() ends up going through sqrt_var(). Also, the
> changes are local to functions in numeric.c, which makes them easy to
> revert if something proves to be wrong.

FWIW, I agree that this is a reasonable thing to consider committing
for v13. It's not adding any new user-visible behavior, so there's
no definitional issues to quibble over, which is usually what I worry
about regretting after an overly-hasty commit. And it's only touching
a few functions in one file, so even if the patch is a bit long, the
complexity seems pretty well controlled.

I've not read the patch in detail so this isn't meant as a review,
but from a process standpoint I see no reason not to go forward.

regards, tom lane