Re: Document atthasmissing default optimization avoids verification table scan

From: Robert Haas <robertmhaas(at)gmail(dot)com>
To: James Coleman <jtc331(at)gmail(dot)com>
Cc: "David G(dot) Johnston" <david(dot)g(dot)johnston(at)gmail(dot)com>, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Andrew Dunstan <andrew(at)dunslane(dot)net>, "Bossart, Nathan" <bossartn(at)amazon(dot)com>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Document atthasmissing default optimization avoids verification table scan
Date: 2022-03-27 15:43:45
Message-ID: CA+Tgmoa2RQdy1uivN2K6JjHQ1PMzkP5qPjcc-qS1T=7Qm+WUUg@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Sat, Mar 26, 2022 at 6:25 PM James Coleman <jtc331(at)gmail(dot)com> wrote:
> I simply do not accept the claim that this is not a reasonable concern
> to have nor that this isn't worth documenting.

I don't think I said that the concern wasn't reasonable, but I don't
think the fact that one person or organization had a concern means
that it has to be worth documenting. And I didn't say either that it's
not intrinsically worth documenting. I said it doesn't fit nicely into
the documentation we have.

Since you didn't like my last example, let's try another one. If
someone shows up and proposes a documentation patch to explain what a
BitmapOr node means, we're probably going to reject it, because it
makes no sense to document that one node and not all the others. That
doesn't mean that people shouldn't want to know what BitmapOr means,
but it's just not sensible to document that one thing in isolation,
even if somebody somewhere happened to be confused by that thing and
not any of the other nodes.

In the same way, I think you're trying to jam documentation of one
particular point into the documentation when there are many other
similar points that are not documented, and I think it's very awkward.
It looks to me like you want to document that a table scan isn't
performed in a certain case when we haven't documented the rule that
would cause that table scan to be performed in other cases, or even
what a table scan means in this context, or any of the similar things
that are equally important, like a table rewrite or an index rebuild,
or any of the rules for when those things happen.

It's arguable in my mind whether it is worth documenting all of those
rules, although I am not opposed to it if somebody wants to do the
work. But I *am* opposed to documenting that a certain situation is an
exception to an undocumented rule about an undocumented concept.
That's going to create confusion, not dispel it.

--
Robert Haas
EDB: http://www.enterprisedb.com

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2022-03-27 15:53:58 Re: Add a pg_get_query_def function (was Re: Deparsing rewritten query)
Previous Message Robert Haas 2022-03-27 14:24:14 Re: Assert in pageinspect with NULL pages