Hi! Amit, on previous email, patch #2 - I agree that it is not the best idea to introduce new type of logic into the parser, so this logic could be moved to the executor, or removed at all. What do you think of these options?
On Wed, Oct 18, 2023 at 5:19 AM jian he <jian.universal...@gmail.com> wrote: > Hi. > based on v22. > > I added some tests again json_value for the sake of coverager test. > > A previous email thread mentioned needing to check *empty in > ExecEvalJsonExpr. > since JSON_VALUE_OP, JSON_QUERY_OP, JSON_EXISTS_OP all need to have > *empty cases, So I refactored a little bit. > might be helpful. Maybe we can also refactor *error cases. > > The following part is not easy to understand. > res = ExecPrepareJsonItemCoercion(jbv, > + jsestate->item_jcstates, > + &post_eval->jcstate); > + if (post_eval->jcstate && > + post_eval->jcstate->coercion && > + (post_eval->jcstate->coercion->via_io || > + post_eval->jcstate->coercion->via_populate)) > -- Regards, Nikita Malakhov Postgres Professional The Russian Postgres Company https://postgrespro.ru/
