Re: scram_iterations is undocumented GUC_REPORT

> On 30 Jan 2024, at 13:36, Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org> wrote:
>
> I noticed while answering a question that commit b577743000cd added the
> GUC scram_iterations and marked it GUC_REPORT, but failed to add it to
> the PQparameterStatus documentation.

Ugh, thanks for fixing!

> 1. that list looks to be in random order. Should we sort it
> alphabetically?

It seems to have some semblance of grouping per GUC functionality, but not
quite. I'm not sure sorting alphabetically will improve the current state
though.

> 2. the notes about the versions in which some parameters started to be
> reported, look quite outdated. We don't really care about things not
> reported in 8.0 or 8.1 or even 9.0. For all purposes, it seems
> perfectly OK to say that these parameters have been reported forever
> (i.e. don't mention them in these lists). I think we should remove all
> those, except the note about version 14.

Agreed.

> 3. Should we list scram_iterations as having started to be reported with
> version 16?

Yes, similar to in_hot_standby for v14.

> The GUC didn't exist before that; but we could say that if
> it's not reported, then the application can assume that the value is
> 4096 (similar to the wording for standard_conforming_strings).

There is no real practical use for knowing the value if it's not reported,
since there isn't anything the user can do differently knowing that. I would
leave that out to avoid confusion, but I don't have strong feelings if you
think it should be added.

--
Daniel Gustafsson