Bug#506833: Scilab fails VERY SIMPLE computation.

[KOBAYASHI Hiroyuki]

Package: scilab
Version: 4.1.2-6


Scilab mistakes computation of (a certain type of?) product of 2 matrices.
Here is an example:

--> A = [3 1 4; 1 5 9; 2 6 5];

--> inv(A)*A
 ans  =
 
    2.  - 8.327D-17  - 1.804D-16  
    0.    2.           0.         
    0.    0.           1.         
 

In Windows, it results with an identity matrix successfully.

I am using a several Debian GNU/Linux IA32 PCs.  Their kernels are
2.6.26.1-1-686 or higher and libc6s are 2.7-16.



-- System Information:
Debian Release: lenny/sid
  APT prefers testing
  APT policy: (500, 'testing'), (500, 'stable')
Architecture: i386 (i686)

Kernel: Linux 2.6.26-1-686 (SMP w/2 CPU cores)
Locale: LANG=ja_JP.EUC-JP, LC_CTYPE=ja_JP.EUC-JP (charmap=EUC-JP)
Shell: /bin/sh linked to /bin/bash

Versions of packages scilab depends on:
ii  scilab-bin                    4.1.2-6    Matrix-based scientific software p
ii  tk8.4 [wish]                  8.4.19-2   Tk toolkit for Tcl and X11, v8.4 -

Versions of packages scilab recommends:
ii  scilab-doc                    4.1.2-6    Matrix-based scientific software p

Versions of packages scilab suggests:
pn  gfortran                      <none>     (no description available)

-- no debconf information
----
KOBAYASHI, Hiroyuki