Re: Cleaning up array_in()

Hello Jian,

05.09.2023 02:53, jian he wrote:
> On Mon, Sep 4, 2023 at 8:00 AM jian he <jian(dot)universality(at)gmail(dot)com> wrote:
>> hi.
>> attached v4.
>> v4, 0001 to 0005 is the same as v3 in
>> https://www.postgresql.org/message-id/5859ce4e-2be4-92b0-c85c-e1e24eab57c6%40iki.fi
>>
>> v4-0006 doing some modifications to address the corner case mentioned
>> in the previous thread (like select '{{1,},{1},}'::text[]).
>> also fixed all these FIXME, Heikki mentioned in the code.
>>
>> v4-0007 refactor ReadDimensionInt. to make the array dimension bound
>> variables within the INT_MIN and INT_MAX. so it will make select
>> '[21474836488:21474836489]={1,2}'::int[]; fail.
>
> attached, same as v4, but delete unused variable {nitems_according_to_dims}.

Please look at the differences, I've observed with the latest patches
applied, old vs new behavior:

Case 1:
SELECT '{1,'::integer[];
ERROR:  malformed array literal: "{1,"
LINE 1: SELECT '{1,'::integer[];
               ^
DETAIL:  Unexpected end of input.

vs

ERROR:  malformed array literal: "{1,"
LINE 1: SELECT '{1,'::integer[];
               ^

(no DETAIL)

Case 2:
SELECT '{{},}'::text[];
ERROR:  malformed array literal: "{{},}"
LINE 1: SELECT '{{},}'::text[];
               ^
DETAIL:  Unexpected "}" character

vs
 text
------
 {}
(1 row)

Case 3:
select '{\{}'::text[];
 text
-------
 {"{"}
(1 row)

vs
 text
------
 {""}

Best regards,
Alexander