Re: left-right (up-down) parsing may lead to over/underflow of prod

[David Bateman]

Rolf Fabian wrote:
> What is actually your intention concerning the development of Octave ?
>
> It seems to me that perfect replication of MatLab's bugs which even severly 
> contradict basic laws of algebra have higher precedence for
> you than getting correct results.
>
>   
Not at all.. I implemented speye(m).^0, etc mathematically correctly
prior to matlab also doing so in 2007a.. I was pointing out that matlab
has the same behavior and therefore we need to be careful if we
implement anything else as differences in behavior result in bug reports
as well. Note the interesting discussion recently on the mldivide
differences between matlab and octave for rank deficient matrices where
the final word is Octave has it right and we shouldn't go the matlab way
just to gain a bit of speed.

D.




-- 
David Bateman                                address@hidden
Motorola Labs - Paris                        +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 6 72 01 06 33 (Mob) 
91193 Gif-Sur-Yvette FRANCE                  +33 1 69 35 77 01 (Fax) 

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary