Re: BUG #8870: PL/PgSQL, SELECT .. INTO and the number of result columns

* marko(at)joh(dot)to (marko(at)joh(dot)to) wrote:
> I looked more closely into a gripe from Stephen Frost, which lead me into
> discovering this behaviour:

Well, as I got mentioned here..

The specific gripe I was bitching about on IRC goes thusly:

=*# do $$
declare f1 int;
declare f2 int;
declare f3 int;
begin
select
1,
2,
3
into
f1,
f2
f3;
end
$$ language plpgsql;
DO

The issue above being that when you have a long list of columns being
selected into variables, it can be easy to forget a comma and then
debugging can be very painful.

> Which directly contradicts this statement in the docs
> (http://www.postgresql.org/docs/9.3/static/plpgsql-statements.html#PLPGSQL-STATEMENTS-SQL-ONEROW)
>
> "If a row or a variable list is used as target, the query's result columns
> must exactly match the structure of the target as to number and data types,
> or else a run-time error occurs."

Agreed.

> This seems to have been broken by commit
> 8bb3c8fe5413103c30ba8d1bcb3a1f8e46bddf3d. I strongly believe the
> documentation is right in this case, and the behaviour ought to change.

Yeah, I don't entirely follow the logic in that commit. It's strongly
discouraged to use 'SELECT * INTO', for pretty obvious reasons. We
certainly have no sympathy for users using 'SELECT *' from other
languages when the columns get rearranged underneath them (due to a DROP
COLUMN or similar).

Thanks,

Stephen