[FFmpeg-devel] [PATCH][RFC] Lagarith Decoder.

[Reimar Döffinger]

On Thu, Sep 17, 2009 at 05:32:46PM +0100, M?ns Rullg?rd wrote:
> Reimar D?ffinger <Reimar.Doeffinger at gmx.de> writes:
> > It can force rounding to IEEE precision of intermediate results, or a
> > specific evaluation order.
> 
> Can != does.  This is not required or even documented behaviour.  We
> should not rely on it.

Well, sure it is not reliable, after all we can not even know which
rounding mode will be used for float operations, can we (haven't checked
the C standard)? Sure there are functions to get and set that, but
FFmpeg does not use them.
Still, it might at least help find out where exactly things go wrong.

[...]
> I think the proper solution is something that avoids floating-point
> entirely.

For which you'd still first have to find out what the "right" behaviour
is, as such "fixing" (as in making it work right at least with one
compiler on one architecture) first might be helpful.