Re: Patch for zeros() and ones()

[John W. Eaton]

On  6-Aug-2003, Heine Kolltveit <address@hidden> wrote:

| On Wed, 30 Jul 2003, Heine Kolltveit wrote:
| > This patch extends zeros() and ones() to be able to create nd-arrays.
| > Thanks to Roderick Köehle for providing much of the code for this one.
| 
| Here is the re-formatted patch.

Thanks for the patch.  I think we are starting to make some progress.

I'm not sure what happened, but the patch for data.cc did not apply
cleanly.  I also made a few changes to check error_state in a few more
places and also to create the octave_value object using

  retval =  ArrayN<double> (dims, val);
  
instead of

  retval =  new octave_double_nd_array (ArrayN<double> (dims, val));

so that the proper constructor for the octave_value object is called
(otherwise, it doesn't get the chance to perform the narrowing
operation).

You should probably update and take a look at what I did.

| 2003-08-06  Heine Kolltveit  <address@hidden>
| 
|       *utils.h (check_dimensions(Array<int>&, char) and
|       (get_dimensions(octave_value&, char, Array<int>&) and
|       *utils.cc (check_dimensions(Array<int>&, char) and
|       (get_dimensions(octave_value&, char, Array<int>&): new
|       *data.cc (fill_matrix(...)): Added functionality to create nd-arrays
|       (DEFUN(ones,...) and (DEFUN(zeros,...): Added the new syntax.

Please take a look at the way other ChangeLog entries are formatted
and follow the conventions already used.

Instead of DEFUN(ones,...), just write (Fones).  This is a convention
used in Octave ChangeLogs (and copied from Emacs) when referring to
built-in functions declared with DEFUN (which actually defines a
function with the name prefixed with the character F).

Finally, can we agree to call these objects "N-d arrays"?

Thanks,

jwe