RE: Big Performance drop of Exceptions in UDFs between V11.2 and 13.4

> -----Original Message-----
> From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
> Sent: Friday, August 27, 2021 13:43
> To: ldh(at)laurent-hasson(dot)com
> Cc: Justin Pryzby <pryzby(at)telsasoft(dot)com>; Ranier Vilela
> <ranier(dot)vf(at)gmail(dot)com>; pgsql-performance(at)postgresql(dot)org
> Subject: Re: Big Performance drop of Exceptions in UDFs between V11.2
> and 13.4
>
> "ldh(at)laurent-hasson(dot)com" <ldh(at)laurent-hasson(dot)com> writes:
> > That being said, do you have any suggestion how I could circumvent
> the
> > issue altogether?
>
> Based on Andrew's report, it seems like you might be able to work
> around it for the time being by disabling message translations, i.e.
> SET lc_messages = 'C';
> Even if that's not acceptable in your work environment, it would be
> useful to verify that you see an improvement from it.
>
> regards, tom lane

Hello Tom.... hit the send button accidentally.

SET lc_messages = 'C';
drop table sampletest;
create table sampletest (a varchar, b varchar);
insert into sampletest (a, b)
select substr(md5(random()::text), 0, 15), (100000000*random())::integer::varchar
from generate_series(1,100000);

CREATE OR REPLACE FUNCTION toFloat(str varchar, val real)
RETURNS real AS $$
BEGIN
RETURN case when str is null then val else str::real end;
EXCEPTION WHEN OTHERS THEN
RETURN val;
END;
$$ LANGUAGE plpgsql COST 1 IMMUTABLE;

show lc_messages; --> OK 'C'

explain (analyze,buffers,COSTS,TIMING)
select MAX(toFloat(b, null)) as "b" from sampletest

Aggregate (cost=2137.00..2137.01 rows=1 width=4) (actual time=175.551..175.552 rows=1 loops=1)
Buffers: shared hit=637
-> Seq Scan on sampletest (cost=0.00..1637.00 rows=100000 width=8) (actual time=0.014..9.270 rows=100000 loops=1)
Buffers: shared hit=637
Planning Time: 0.087 ms
Execution Time: 175.600 ms

explain (analyze,buffers,COSTS,TIMING)
select MAX(toFloat(a, null)) as "a" from sampletest

Aggregate (cost=2137.00..2137.01 rows=1 width=4) (actual time=88031.549..88031.551 rows=1 loops=1)
Buffers: shared hit=637
-> Seq Scan on sampletest (cost=0.00..1637.00 rows=100000 width=15) (actual time=0.008..34.494 rows=100000 loops=1)
Buffers: shared hit=637
Planning:
Buffers: shared hit=4
Planning Time: 0.171 ms
Execution Time: 88031.585 ms

Doesn't seem to make a difference unless I misunderstood what you were asking for regarding the locale?

Thank you,
Laurent.