pg.ddx.io  pgsql-bugs@postgresql.org mailing list archive  
help / color / mirror / Atom feed
From: PG Bug reporting form <noreply@postgresql.org>
To: pgsql-bugs@lists.postgresql.org
Cc: 303677365@qq.com
Subject: BUG #19811: numeric log(base, x): wrong sign of Infinity result when base < 1; base 1 accepted with x = Infinity
Date: Thu, 29 Oct 2026 06:37:56 +0000
Message-ID: <19811-3e5c60ea0ef2a9b7@postgresql.org> (raw)

The following bug has been logged on the website:

Bug reference:      19811
Logged by:          chunling qin
Email address:      303677365@qq.com
PostgreSQL version: 18.6
Operating system:   Linux x86_64
Description:        

The two-argument numeric logarithm returns +Infinity for x = Infinity
regardless of the base. For bases between 0 and 1 the correct result is
-Infinity, and for base 1 the call should be rejected.

Self-contained test (official postgres:latest docker image, 18.6):

    SELECT log('0.5'::numeric, '1e308'::numeric);     --
-1023.15385322530760   (correct sign)
    SELECT log('0.5'::numeric, 'Infinity'::numeric);  -- Infinity   (wrong:
should be -Infinity)
    SELECT log('0.1'::numeric, 'Infinity'::numeric);  -- Infinity   (wrong)
    SELECT log('0.25'::numeric, 'Infinity'::numeric); -- Infinity   (wrong)

Controls:

    SELECT log('2'::numeric, 'Infinity'::numeric);    -- Infinity
(correct, base > 1)
    SELECT ln('Infinity'::numeric);                   -- Infinity
(correct)
    SELECT log('0.5'::numeric, '-Infinity'::numeric); -- ERROR: cannot take
logarithm of a negative number

The function is internally inconsistent: with a large finite argument the
sign is correct (-1023 for 1e308, since ln(x)/ln(0.5) is negative), but at
exactly Infinity the sign flips to positive. Mathematically log_b(Infinity)
= ln(Infinity)/ln(b) = Infinity/ln(b), which is -Infinity for 0 < b < 1 and
+Infinity for b > 1.

Base 1 is also not rejected when x = Infinity:

    SELECT log('1'::numeric, '12.34'::numeric);       -- ERROR: division by
zero    (correct rejection)
    SELECT log('1'::numeric, 'Infinity'::numeric);    -- Infinity
(wrong: the same call with a finite x is rejected)

The Infinity special case in the numeric log implementation appears to
shortcut the ln(x)/ln(b) computation without considering the sign of ln(b),
or the ln(b) = 0 case.








view thread (2+ messages)  latest in thread

Message-ID: <19811-3e5c60ea0ef2a9b7@postgresql.org>
Permalink:  ../19811-3e5c60ea0ef2a9b7@postgresql.org/
Also on:    postgresql.org/message-id/19811-3e5c60ea0ef2a9b7@postgresql.org

 · 

reply

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Reply to all the recipients from your mail client; the link fills in
  the headers below: reply via email
  Some clients, webmail especially, drop In-Reply-To and References from
  a link, and the reply then starts a new thread: use the mbox below.

  To: pgsql-bugs@postgresql.org
  Cc: noreply@postgresql.org, 303677365@qq.com
  Subject: Re: BUG #19811: numeric log(base, x): wrong sign of Infinity result when base < 1; base 1 accepted with x = Infinity
  In-Reply-To: <19811-3e5c60ea0ef2a9b7@postgresql.org>
  References: <19811-3e5c60ea0ef2a9b7@postgresql.org>

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox