Re: Octave 2.9.6 available for ftp [files]

[David Bateman]

Thomas Weber wrote:

>Ok, second try, this time with files included :)
>
>Sorry for the noise.
>  
>
I'm sending this back to the maintainers list so others can comment. The
error in the rand I see in the log files is

error: assert (mean (x),a,0.03) expected
 5
but got
 5.0576
maximum absolute error 0.05761 exceeds tolerance 0.03

so it seems to me that this really is just a statistical glitch. Perhaps
Bill's idea of fixing the seed/state would address this.

As for the three errors in the sparse code, they are all

error: assert (bf .\ as,sparse (bf .\ af, true),100 * eps) expected
   Inf +  Infi     2 -    1i   NaN -  NaNi   NaN -  NaNi
     0 +    0i   NaN -  NaNi   NaN -  NaNi   Inf +  Infi
     0 +    0i     0 +    0i   NaN -  NaNi   Inf -  NaNi
but got
   Inf +  Infi     2 -    1i   NaN -  NaNi   NaN -  NaNi
     0 +    0i   NaN -  NaNi   NaN -  NaNi   Inf +  Infi
     0 +    0i     0 +    0i   NaN -  NaNi   Inf -  NaNi
Infs don't match

Looking at the above the Inf's do match. Probably the 4.1.x compiler is
using a  difference representation of Inf in IEE754 for each of these
cases. So it seems to me that it is a gcc bug as it doesn't consistently
use the same Inf representation.

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