Обсуждение: BUG #3289: SIN(PI()) expected to return 0, but returns garbage

The following bug has been logged online:

Bug reference:      3289
Logged by:          Daniel Kastenholz
Email address:      mail@danielkastenholz.de
PostgreSQL version: 8.1, 8.2.4
Operating system:   Linux, Windows
Description:        SIN(PI()) expected to return 0, but returns garbage
Details:

Trouble case:

Action:
Type in SELECT SIN(PI())

Expected output:
0

Actual output:
* Windows, using 8.2.4:
  garbage (-2.2......)

* Linux, using 8.1:
  garbage (1.22.......)

--

Comments:
SIN(3.1414) and SIN(3.1417) come close to 0 and work on both platforms.
SIN(3.1415) and SIN(3.1416) produce the same garbage output as using PI() on
both platforms.

Daniel Kastenholz wrote:
> Type in SELECT SIN(PI())
>
> Expected output:
> 0
>
> Actual output:
> * Windows, using 8.2.4:
>   garbage (-2.2......)
>
> * Linux, using 8.1:
>   garbage (1.22.......)
>
> --
>
> Comments:
> SIN(3.1414) and SIN(3.1417) come close to 0 and work on both platforms.
> SIN(3.1415) and SIN(3.1416) produce the same garbage output as using PI() on
> both platforms.

You don't get 0 because the value of pi is just an approximation, and
most likely the output of sin is not exact either. What you get is
pretty darn close to 0, however. Apparently you didn't read the output
carefully: it's in scientific notation. I'm getting

  1.22460635382238e-16

on my laptop, which means 1.22460635382238 * 10^(-16).

--
   Heikki Linnakangas
   EnterpriseDB   http://www.enterprisedb.com