Re: ALTER TYPE OWNER fails to recurse to multirange

Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> On Tue, Jan 16, 2024 at 11:46 AM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> Also, we already
>> treat the multirange as dependent for some things:

> But this seems like an entirely valid point.

Yeah, it's a bit of a muddle. But there is no syntax for making
a standalone multirange type, so it seems to me that we've mostly
determined that multiranges are dependent types. There are just
a few places that didn't get the word.

Attached is a proposed patch to enforce that ownership and permissions
of a multirange are those of the underlying range type, in ways
parallel to how we treat array types. This is all that I found by
looking for calls to IsTrueArrayType(). It's possible that there's
some dependent-type behavior somewhere that isn't conditioned on that,
but I can't think of a good way to search.

If we don't do this, then we need some hacking in pg_dump to get it
to save and restore these properties of multiranges, so it's not like
the status quo is acceptable.

I'd initially thought that perhaps we could back-patch parts of this,
but now I'm not sure; it seems like these behaviors are a bit
intertwined. Given the lack of field complaints I'm inclined to leave
things alone in the back branches.

regards, tom lane