Re: GUC values - recommended way to declare the C variables?

On Thu, Oct 27, 2022 at 11:33:48AM +0900, Michael Paquier wrote:
> On Wed, Oct 26, 2022 at 09:14:37PM -0500, Justin Pryzby wrote:
> > It seems like you're reviewing the previous version of the patch, rather
> > than the one attached to the message you responded to (which doesn't
> > have anything to do with GUC_DEFAULT_COMPILE).
>
> It does not seem so as things stand, I have been looking at
> v5-0001-GUC-C-variable-sanity-check.patch as posted by Peter here:
> https://www.postgresql.org/message-id/CAHut+PuCHjYXiTGdTOvHvDnjpbivLLr49gWVS+8VwnfoM4hJTw@mail.gmail.com

This thread is about consistency of the global variables with what's set
by the GUC infrastructure.

In v4, Peter posted a 2-patch series with my patch as 001.
But I pointed out that it's better to fix the initialization of the
compile-time GUCs rather than exclude them from the check.
Then Peter submitted v5 whcih does that, and isn't built on top of my
patch.

> In combination with a two-patch set as posted by you here:
> 0001-add-DYNAMIC_DEFAULT-for-settings-which-vary-by-.-con.patch
> 0002-WIP-test-guc-default-values.patch
> https://www.postgresql.org/message-id/20221024220544.GJ16921@telsasoft.com

That's a separate thread regarding consistency of the default values
(annotations) shown in postgresql.conf. (I'm not sure whether or not my
patch adding GUC flags is an agreed way forward, although they might
turn out to be useful for other purposes).

--
Justin