Re: [PROPOSAL] new diagnostic items for the dynamic sql

po 8. 11. 2021 v 5:24 odesílatel Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com>
napsal:

>
>
> po 8. 11. 2021 v 5:07 odesílatel Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com>
> napsal:
>
>>
>>>
>>> +set_errcurrent_query (const char *query)
>>>
>>> You can remove the space prior to (.
>>> I wonder if the new field can be named current_err_query because that's
>>> what the setter implies.
>>> current_query may give the impression that the field can store normal
>>> query (which doesn't cause exception).
>>> The following code implies that only one of internalquery and
>>> current_query would be set.
>>>
>>
>> yes, I think so current_query is not a good name too. Maybe query can be
>> good enough - all in ErrorData is related to error
>>
>
> so the name of field can be query, and routine for setting errquery or
> set_errquery
>

and this part is not correct

<--><-->switch (carg->mode)
<--><-->{
<--><--><-->case RAW_PARSE_PLPGSQL_EXPR:
<--><--><--><-->errcontext("SQL expression \"%s\"", query);
<--><--><--><-->break;
<--><--><-->case RAW_PARSE_PLPGSQL_ASSIGN1:
<--><--><-->case RAW_PARSE_PLPGSQL_ASSIGN2:
<--><--><-->case RAW_PARSE_PLPGSQL_ASSIGN3:
<--><--><--><-->errcontext("PL/pgSQL assignment \"%s\"", query);
<--><--><--><-->break;
<--><--><-->default:
<--><--><--><-->set_errcurrent_query(query);
<--><--><--><-->errcontext("SQL statement \"%s\"", query);
<--><--><--><-->break;
<--><-->}
<-->}

set_errcurrent_query should be outside the switch

We want PG_SQL_TEXT for assign statements too

_t := (select ...);

Regards

Pavel