On 9/21/25 19:34, Tom Lane wrote:
> Jim Jones <jim.jones@uni-muenster.de> writes:
>> That's indeed a much larger problem. Calling it from a session silently
>> delivers a "wrong" result --- I was expecting an error.
>
> Yeah, me too. See
>
> https://www.postgresql.org/message-id/2736425.1758475979%40sss.pgh.pa.us
>
The attached PoC now raises an ERROR instead of a WARNING.
A boolean is now computed in fmgr_sql_validator(), set to true if the
function has a prosqlbody (BEGIN ATOMIC) and is defined in a
non-temporary schema. This flag is then used to call
check_sql_fn_statements().
In check_sql_fn_statements(): if the new flag is true, it scans the
function body and raises an error if any temporary relations are found;
if it's false, it skips that check.
In returning.sql there was a query that creates a BEGIN ATOMIC function
using on a temporary table. I changed the table to permanent.
Best regards, Jim
