On Wed, Nov 28, 2018 at 9:17 PM Stephen Frost <sfrost@snowman.net> wrote:
> > Actually, some geometric comparisons are performed counting
> > tolerance margin, the validity of which is in doubt. Their
> > behavior has been changed in recent major version and still has a
> > room for improvement, and the functions are parallel-safe and
> > immutable. Immutablity is mentiond mainly in the light of
> > optimization in the documentation.
>
> I really don't buy off on these arguments in the least. I also didn't
> say that a function wasn't allowed to change- but that the output of an
> immutable function, for a given input, shouldn't change and in the very
> rare case where we absolutely had to make a change, it had better be for
> a very good reason and we need to consider the impact on user indexes.
I think you're getting a little carried away here. Kyotaro-san's
argument seems extremely strong to me, so much so that I can't really
see how you can argue against it. If we have a bug that says 1 + 1 =
3, we are going to fix it. We're not going say, oh, well, it's an
immutable function, so we're just going to carry on having it return
the wrong answer. That would be ridiculous.
It feels to me like you are trying to turn this into some kind of
principled stand against evil people who don't care about
immutability, but rumors that the barbarians are at the gates seem
greatly exaggerated from where I sit.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
