Re: BUG in postgres mathematic

On Thursday 25 January 2001 22:52, Tom Lane wrote:
> "Robert B. Easter" <reaster@comptechnews.com> writes:
> > This problem is not specific to Postgres.
>
> The fact that 5*27.81*100 != 27.81*100*5 is certainly a garden-variety
> floating-point roundoff error.  However, I think Max has a fair
> complaint here: it seems float-to-int8 conversion is truncating, not
> rounding like the other conversions to integer do.
>
> regression=# select 4.7::float8::int4;
>  ?column?
> ----------
>         5
> (1 row)
>
> regression=# select 4.7::float8::int8;
>  ?column?
> ----------
>         4
> (1 row)
>
> Seems to me this is a bug we should fix.
>
>             regards, tom lane

Yeah, I agree. It isn't right that it truncates and that is something C does
appearently.  The fix is to pass the float through a rounding something like
(long)(f + 0.5) or else C just truncates it off.  This must already be
happening for the int4 conversion or C would do the same thing to it.  I
didn't look at the Postgres sources yet, but it is probably one of those very
easy things to fix. :)


--
-------- Robert B. Easter  reaster@comptechnews.com ---------
-- CompTechNews Message Board http://www.comptechnews.com/ --
-- CompTechServ Tech Services http://www.comptechserv.com/ --
---------- http://www.comptechnews.com/~reaster/ ------------