Re: Document "59.2. Built-in Operator Classes" have a clerical error?

On Sun, Aug  2, 2020 at 08:43:53PM -0700, David G. Johnston wrote:
> On Sun, Aug 2, 2020 at 8:17 PM osdba <mailtch@163.com> wrote:
> 
>     hi all:
> 
>     In Document "Table 59-1. Built-in GiST Operator Classes":
> 
>     "range_ops any range type && &> &< >> << <@ -|- = @> @>"， exist double "@>
>     "，
>      
>     Should be "<@ @>" ?
> 
> 
> 
> It helps to reference the current version of the page (or provide a url link)
> as that section seems to have migrated to Chapter 64 - though it is unchanged
> even on the main development branch.
> 
> The table itself is extremely difficult to read: it would be more easily
> readable if the font was monospaced, but its not.
> 
> I'm reasonably confident that the equal sign is part of the second-to-last
> operator while the lone @> is the final operator.  Mostly I say this because
> GiST doesn't do straight equality so a lone equal operator isn't valid.

I dug into this.  This query I think explains why the duplicate is
there:

    SELECT oprname, oprleft::regtype, oprright::regtype, oprresult::regtype
    FROM pg_am
        JOIN pg_opclass ON opcmethod = pg_am.oid
        JOIN pg_amop ON opcfamily = pg_amop.amopfamily
        JOIN pg_operator ON amopopr = pg_operator.oid
    WHERE amname = 'gist'
        AND opcname = 'range_ops'
    ORDER BY 1

     oprname | oprleft  |  oprright  | oprresult
    ---------+----------+------------+-----------
     &&      | anyrange | anyrange   | boolean
     &<      | anyrange | anyrange   | boolean
     &>      | anyrange | anyrange   | boolean
     -|-     | anyrange | anyrange   | boolean
     <<      | anyrange | anyrange   | boolean
     <@      | anyrange | anyrange   | boolean
     =       | anyrange | anyrange   | boolean
     >>      | anyrange | anyrange   | boolean
-->     @>      | anyrange | anyrange   | boolean
-->     @>      | anyrange | anyelement | boolean

Notice that @> appears twice.  (I am not sure why @> appears twice in
the SQL output, while <@ appears only once.)  The PG docs explain the
duplicate:

    https://www.postgresql.org/docs/12/functions-range.html

    @>     contains range     int4range(2,4) @> int4range(2,3)     t
    @>     contains element     '[2011-01-01,2011-03-01)'::tsrange @> '2011-01-10'::timestamp     t
    <@     range is contained by     int4range(2,4) <@ int4range(1,7)     t
    <@     element is contained by     42 <@ int4range(1,7)     f

There is an anyrange/anyrange version, and an anyrange/anyelement
version of @> and <@.  Anyway, for the docs, I think we can either
remove the duplicate entry, or modify it to clarify one is for
anyrange/anyrange and another is for anyrange/anyelement.  I suggest the
first option.

-- 
  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
  EnterpriseDB                             https://enterprisedb.com

  The usefulness of a cup is in its emptiness, Bruce Lee