Re: How to use the OO features of Octave

[David Bateman]

David Bateman wrote:
> David Bateman wrote:
> > John W. Eaton wrote:
> > >
> > >      args(2) = num_indices;
> > >      args(1) = index_position;
> > >      args(0) = *indexed_object;
> >
> >
> > shouldn't args(1) be
> >
> > args(1) = index_position + 1;
> >
> > due to the 1-based indexing of Octave?
> >
> > D.
> Also shouldn't we subtract one from the return value of the feval call?
>
> D.

Ok, I see the return value should be minus one, but index_pos definitely 
needs to be. As far as I can see a pretty generic NDArray end method 
would look like


function r = end (obj, index_pos, num_indices)
  ++index_pos;
  dv = size (obj.x);
  for i = num_indices + 1 : length (dv)
    dv(num_indices) *= dv(i);
  endfor
  if (index_pos < length (dv))
    r = dv (index_pos);
  elseif
    r = 1;
  endif
endfunction


though the "++index_pos;" should go away if F__end__ was adapted to pass 
"index_pos + 1". However, this doesn't work correctly with the toy class 
What I see is

a = gf (floor(4*rand(3,3,2)),2);

a(1,1:end)

ans =
GF(2^2) array. Primitive Polynomial = D^2+D+1 (decimal 7)

Array elements =

0

and

a(1,1,1:end)

ans =
GF(2^2) array. Primitive Polynomial = D^2+D+1 (decimal 7)

Array elements =

0

Not sure what is going wrong but the first of the above should produce 6 
elements and the second 2 elements, but both produce 1. Any ideas?

On a related note "a(:)" on the class object works correctly, but 
"a(1,:)" and "a(:,1)" etc don't

D.