Re: Document atthasmissing default optimization avoids verification table scan

On Fri, Mar 25, 2022 at 5:00 PM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>
> Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> > I vote for rejecting both of these patches.
>
> I see what James is on about here, but I agree that these specific changes
> don't help much. What would actually be desirable IMO is a separate
> section somewhere explaining the performance characteristics of ALTER
> TABLE. (We've also kicked around the idea of EXPLAIN for ALTER TABLE,
> but that's a lot more work.) This could coalesce the parenthetical
> remarks that exist in ddl.sgml as well as alter_table.sgml into
> something a bit more unified and perhaps easier to follow. In particular,
> it should start by defining what we mean by "table rewrite" and "table
> scan". I don't recall at the moment whether we define those in multiple
> places or not at all, but as things stand any such discussion would be
> pretty fragmented.
>
> regards, tom lane

I think a unified area discussing pitfalls/performance of ALTER TABLE
seems like a great idea.

That being said: given that "as things stand" that "discussion
[already is] pretty fragmented" is there a place for a simpler
improvement (adding a short explanation of this particular hazard) in
the meantime? I don't mean this specific v4 patch -- just in general
(since the patch can be revised of course).

Thanks,
James Coleman