Re: Dispatch function

[John W. Eaton]

On 05/14/2016 12:38 PM, pritika malhotra wrote:
> On 14 May 2016 at 21:54, Julien Bect <address@hidden> wrote:
> > Le 14/05/2016 17:18, pritika malhotra a écrit :
> > > Hello! I was trying to use this given example from documentation and as this 
> > > function is no longer available.
> > > Then what will be the replacement of "dispatch ("sin", "spsin", "sparse 
> > > matrix");"
> > >
> > > Through earlier discussion, I have come to know that we can define classes. How 
> > > can we do that?
> > > A simple illustration if possible Or is there some other to do so.
> > >
> > > function y = spsin (x)
> > >     printf ("Calling spsin\n");
> > >     fflush(stdout);
> > >     y = spfun ("sin", x);
> > > endfunction
> > >
> > > dispatch ("sin", "spsin", "sparse matrix");
> > > y0 = sin(eye(3));
> > > y1 = sin(speye(3));
> >
> >
> > Here is the relevant section of the manual:
> > https://www.gnu.org/software/octave/doc/v4.0.0/Creating-a-Class.html#Creating-a-Class
> >
> > I don't know if you can do it for sparse matrices...
>
> Thanks.
> Yes, I went through this manual.
> So in order to use the same function I need to create class of the same name.
> I just wanted to know if we have any default classes already existing in octave?

Create a directory called @double somewhere in your loadpath and define 
something like the following function there:

  function r = sin (a)
    if (issparse (a))
      r = ...;  ## your special definition for sparse matrices
    else
      r = builtin ('sin', a);
    endif
  endfunction

You can only overload on class name, not the specific types as reported 
by the typeinfo function.  The specific type names like "sparse matrix" 
are not classes in the interpreter.  The class for sparse matrices is 
either double or logical.

Overloading for operators is not yet implemented, so defining a function 
called "plus" won't overload the + operator (for example).

jwe