Re: [AD] [ alleg-Bugs-1650580 ] Problem in fixed point library.

[ Thread Index | Date Index | More lists.liballeg.org/allegro-developers Archives ]


On Mon, 2007-02-05 at 01:18 +1100, Peter Wang wrote:
> On 2007-02-04, Elias Pschernig <elias@xxxxxxxxxx> wrote:
> > I'll apply the below patch if nobody objects - the bug reporter claims
> > to have tested it, and LONG_LONG also is used for fixmul so should cause
> > no problem here.
> > 
> > Index: include/allegro/inline/fmaths.inl
> > ===================================================================
> > --- include/allegro/inline/fmaths.inl   (revision 7706)
> > +++ include/allegro/inline/fmaths.inl   (working copy)
> > @@ -151,7 +151,7 @@
> >        return (x < 0) ? -0x7FFFFFFF : 0x7FFFFFFF;
> >     }
> >     else
> > -      return ftofix(fixtof(x) / fixtof(y));
> > +      return ((LONG_LONG)x << 16) / (LONG_LONG)y;
> >  })
> 
> Didn't we discuss the fixed point functions on allegro.cc last year or
> so?  Don't apply this just yet.

Yes, found it here: http://www.allegro.cc/forums/thread/488085

And thinking about it, there actually is nothing wrong with using float
as implied by the bug report. So, as the float is faster according to
that thread, indeed nothing should be changed then (until someone comes
up with a test to prove the opposite).

-- 
Elias Pschernig





Mail converted by MHonArc 2.6.19+ http://listengine.tuxfamily.org/