Re: Ability to reference other extensions by schema in extension scripts

Sandro Santilli <strk(at)kbt(dot)io> writes:
> On Mon, Mar 13, 2023 at 05:57:57PM -0400, Regina Obe wrote:
>> Attached is a slightly revised patch to fix the extra whitespace in the
>> extend.gml document that Sandro noted to me.

> Thanks Regina.
> I've tested attached patch (md5 0b652a8271fc7e71ed5f712ac162a0ef)
> against current master (hash 4ef1be5a0b676a9f030cc2e4837f4b5650ecb069).
> The patch applies cleanly, builds cleanly, regresses cleanly.

Pushed with some mostly-cosmetic adjustments (in particular I tried
to make the docs and tests neater).

I did not commit the changes in get_available_versions_for_extension
to add no_relocate as an output column. Those were dead code because
you hadn't done anything to connect them up to an actual output parameter
of pg_available_extension_versions(). While I'm not necessarily averse
to making the no_relocate values visible somehow, I'm not convinced that
pg_available_extension_versions should be the place to do it. ISTM what's
relevant is the no_relocate values of *installed* extensions, not those of
potentially-installable extensions. If we had a view over pg_extension
then that might be a place to add this, but we don't. On the whole it
didn't seem important enough to pursue, so I just left it out.

regards, tom lane