rsgenpoly

[Robert T. Short]

    Before I file a bug report and spend the time to fix this, let me 
make sure that I am not misinterpreting the results.  This is regarding 
rsgenpoly from the communications package.

rsgenpoly returns the following from MATLAB:

>> rsgenpoly(31,15)

ans = GF(2^5) array. Primitive polynomial = D^5+D^2+1 (37 decimal)

Array elements =

    1   15   28    1   13    8    2   24   25   20   16   15 29   15   
21    3   14


>> g = rsgenpoly(7,3)

g = GF(2^3) array. Primitive polynomial = D^3+D+1 (11 decimal)

Array elements =

   1   3   1   2   3

octave yields

>> rsgenpoly(31,15)
ans =

 Columns 1 through 9:

   1.0000e+00   2.5700e+02   3.0245e+04   2.1606e+06 1.0470e+08   
3.6442e+09   9.4065e+10   1.8334e+12   2.7211e+13

 Columns 10 through 17:

   3.0784e+14   2.6387e+15   1.6904e+16   7.9057e+16 2.5977e+17   
5.6242e+17   7.1149e+17   3.9190e+17

>> rsgenpoly(7,3)
ans =

     1    15    80   180   144

This is very recent version of MATLAB - either the most recent or the 
one just before.  I have octave 4.0.3 with communications 1.2.1.  I do 
not install the comm packages or any dependencies because there are too 
many conflicts with my own library functions, so I just cherry pick what 
I need and install the bits.

I am pretty sure this is an error and the offending line is a 
convolution that does the arithmetic on the integers instead the 
appropriate Galois field, but I would appreciation some feedback. If it 
is an error I will fix it.  Shouldn't be hard.

DB