Re: [PATCH][Documination] Add optional USING keyword before opclass name in INSERT statemet

On Tue, May 31, 2016 at 3:55 PM, Nikolay Shaplov <n(dot)shaplov(at)postgrespro(dot)ru>
wrote:

> В письме от 31 мая 2016 15:38:38 пользователь Robert Haas написал:
>
> > >>> 99% of the time, you'd be right. But this is an unusual case, for
> the
> > >>> reasons I mentioned before.
> > >>
> > >> I tend to agree with Nikolay. I can't see much upside in making this
> > >> change. At best, nothing will break. At worst, something will break.
> > >> But how do we actually come out ahead?
> > >
> > > We come out ahead by not having to make the documentation more
> confusing.
> > >
> > > Basically, we have the opportunity to fix an ancient mistake here at
> > > very low cost. I do not think that doubling down on the mistake is
> > > a better answer.
> >
> > I'm not convinced, but we don't have to agree on everything...
> I am not convinced too. But I will not argue hard for the patch as far as
> my
> main goal was to report inconsistency. Through the I consider Tom's
> proposal
> quite strange...
>
>
​We've recently chosen to not document the "ANALYZE -> ANALYSE" syntax, and
I'm sure there are other examples, so I don't see why the status quo
(pre-Tom's patch) is unacceptable...if adding USING to the synopsis is
prone to cause confusion then don't; but lets not break existing uses that
in no way harm the project.

Otherwise I presume Tom is correct that the true fix is more than a single
word in one file of our documentation. If you want to see it stay and be
documented there needs to be a complete proposal that at least gets, even
if grudging, approval from a couple of people and a committer.

David J.
​