Re: SQL/JSON features for v15

[Jonathan S. Katz]

On 8/31/22 8:38 AM, Andrew Dunstan wrote:

On 2022-08-31 We 07:01, Amit Langote wrote:

On Wed, Aug 31, 2022 at 3:51 PM Amit Langote <amitlangot...@gmail.com> wrote:

  SELECT JSON_VALUE(jsonb '"aaa"', '$' RETURNING int DEFAULT 111 ON ERROR);
- json_value
-------------
-        111
-(1 row)
-
+ERROR:  syntax error at or near "DEFAULT"
+LINE 1: ...ELECT JSON_VALUE(jsonb '"aaa"', '$' RETURNING int DEFAULT 11...

Is it intentional that you left many instances of the regression test
output changes like the above?

Actually, thinking more about this, I am wondering if we should not
remove the DEFAULT expression productions in gram.y.  Maybe we can
keep the syntax and give an unsupported error during parse-analysis,
like the last version of the patch did for DEFAULT ON EMPTY.  Which
also means to also leave JsonBehavior alone but with default_expr
always NULL for now.

Producing an error in the parse analysis phase seems best to me.

Thanks,

Jonathan

Andres, Robert, Tom: With this recent work, have any of your opinions changed on including SQL/JSON in v15?

OpenPGP_signature
Description: OpenPGP digital signature