Re: Named Operators

On Fri, Jan 20, 2023 at 9:17 AM Gurjeet Singh <gurjeet(at)singh(dot)im> wrote:

> On Sat, Jan 14, 2023 at 6:14 AM Gurjeet Singh <gurjeet(at)singh(dot)im> wrote:
> >
> > I agree that an identifier _surrounded_ by the same token (e.g. #foo#)
> > or the pairing token (e.g. {foo}) looks better aesthetically, so I am
> > okay with any of the following variations of the scheme, as well:
> >
> > \#foo\# (tested; works)
> > \#foo# (not tested; reduces ident length by 1)
> >
> > We can choose a different character, instead of #. Perhaps \{foo} !
>
> Please find attached the patch that uses \{foo} styled Named
> Operators. This is in line with Tom's reluctant hint at possibly using
> curly braces as delimiter characters. Since the curly braces are used
> by the SQL Specification for row pattern recognition, this patch
> proposes escaping the first of the curly braces.
>
> We can get rid of the leading backslash, if (a) we're confident that
> SQL committee will not use curly braces anywhere else, and (b) if
> we're confident that if/when Postgres supports Row Pattern Recognition
> feature, we'll be able to treat curly braces inside the PATTERN clause
> specially. Since both of those conditions are unlikely, I think we
> must settle for the escaped-first-curly-brace style for the naming our
> operators.
>
> Keeping with the previous posts, here's a sample SQL script showing
> what the proposed syntax will look like in action. Personally, I
> prefer the \#foo style, since the \# prefix stands out among the text,
> better than \{..} does, and because # character is a better signal of
> an operator than {.
>
> create operator \{add_point}
> (function = box_add, leftarg = box, rightarg = point);
> create table test(a box);
> insert into test values('((0,0),(1,1))'), ('((0,0),(2,1))');
> select a as original, a \{add_point} '(1,1)' as modified from test;
> drop operator \{add_point}(box, point);
>
> Best regards,
> Gurjeet
> http://Gurje.et

Hi,
Since `validIdentifier` doesn't modify the contents of `name` string, it
seems that there is no need to create `tmp` string in `validNamedOperator`.
You can pass the start and end offsets into the string (name) as second and
third parameters to `validIdentifier`.

Cheers