On 2023/11/20 17:33, Pavel Stehule wrote:
>
>
> I did some deeper check:
>
> - I don't like too much parser's modification (I am sending alternative
> own implementation) - the SQL parser allows richer syntax, and for full
> functionality is only few lines more
Agree.
>
> - original patch doesn't solve %ROWTYPE
>
> (2023-11-20 10:04:36) postgres=# select * from foo;
> ┌────┬────┐
> │ a │ b │
> ╞════╪════╡
> │ 10 │ 20 │
> │ 30 │ 40 │
> └────┴────┘
> (2 rows)
>
> (2023-11-20 10:08:29) postgres=# do $$
> declare v foo%rowtype[];
> begin
> v := array(select row(a,b) from foo);
> raise notice '%', v;
> end;
> $$;
> NOTICE: {"(10,20)","(30,40)"}
> DO
>
two little fixes
1. spelling mistake
ARRAY [ icons ] --> ARRAY [ iconst ]
2. code bug
if (!OidIsValid(dtype->typoid)) --> if (!OidIsValid(array_typeid))
> - original patch doesn't solve type RECORD
> the error message should be more intuitive, although the arrays of
> record type can be supported, but it probably needs bigger research.
>
> (2023-11-20 10:10:34) postgres=# do $$
> declare r record; v r%type[];
> begin
> v := array(select row(a,b) from foo);
> raise notice '%', v;
> end;
> $$;
> ERROR: syntax error at or near "%"
> LINE 2: declare r record; v r%type[];
> ^
> CONTEXT: invalid type name "r%type[]"
>
Currently only scalar variables are supported.
This error is consistent with the r%type error. And record arrays are
not currently supported.
Support for r%type should be considered first. For now, let r%type[]
report the same error as record[].
I prefer to implement it with a new patch.
> - missing documentation
My English is not good. I wrote it down, please correct it. Add a note
in the "Record Types" documentation that arrays and "Copying Types" are
not supported yet.
>
> - I don't like using the word "partitioned" in the regress test name
> "partitioned_table". It is confusing
fixed
>
> Regards
>
> Pavel