Re: PGQ catalog representation and pg_dump support

On Tue, Sep 15, 2026 at 3:18 AM Ashutosh Bapat
<ashutosh(dot)bapat(dot)oss(at)gmail(dot)com> wrote:
> I am mentioning the sections of SQL/PGQ standard which govern the
> behaviour here and showing you that the implementation is compliant
> with the standard. Maybe we missed adding comments mentioning the
> rationale. I have added that as a TODO

Obviously, this is your patch set, and you can proceed how you wish,
but I do not think that just adding some comments is going to be
nearly good enough. The inconsistencies between the rules that had to
hold at CREATE time and what had to hold at ALTER time are very
significant, and that's the sort of thing that usually turns out to
cause problems, as it did here. I don't think we have another DDL
command in the entire system where the gap is as large as what was the
case here, and I don't think a patch that doesn't think harder about
that topic is going to stay committed (or, hopefully, get committed).

--
Robert Haas