В письме от пятница, 8 декабря 2023 г. 15:59:09 MSK пользователь Alvaro
Herrera написал:
> > Theoretically I can create patch with full options.c as it is in the patch
> > now, and use that code only in index AM, and keep reloption.c mostly
> > unchanged.
> >
> > This will be total mess with two different options mechanisms working in
> > the same time, but this might be much more easy to review. When we are
> > done with the first step, we can change the rest.
> > If this will help to finally include patch into postgres, I can do it.
> > Will
> > that help you to review?
>
> I don't think that's better, because we could create slight
> inconsistencies between the code used for index AMs and the users of
> reloptions.
I've written quite good regression tests for it, there should be no
inconsistency.
> I'm not seeing any reasonable way to split this patch in smaller ones.
Actually me neither. Not a good one anyway.
But if somebody really need it to be split, it can be done that way.
--
Nikolay Shaplov aka Nataraj
Fuzzing Engineer at Postgres Professional
Matrix IM: @dhyan:nataraj.su