Re: numel(foo{:}) - feature or bug ?

[Marco atzeri]

On 8/5/2011 6:37 PM, Sergei Steshenko wrote:
> >
>
> Oh, if somebody wants (pseudo)scientific terminology, then here it is:
>
> Why does adding an element causes non-monotonicity in numel(foo2d{:}) ?
>
> Thanks,
>    Sergei.


Sergei,
could you please cool down and trim a bit all your e-mails ?
It is not polite to report every time the full story of your
previous mails, and make following you very hard.

At least 2 persons already tried to explain you that fooo2d
and foo2d{:}) are not the same thing.

You can disagree on how numel works on matlab and octave, but or you
provide a compelling reason for the change and you provide also a
patch for the change, or I doubt numel will be modified.

by the way on GNU Octave, version 3.4.2

> foo
foo =
{
  [1,1] = a
  [2,1] = aa
  [1,2] = b
  [2,2] = [](0x0)
}

> foo(:)
ans =
{
  [1,1] = a
  [2,1] = aa
  [3,1] = b
  [4,1] = [](0x0)
}

> foo{:}
ans = a
ans = aa
ans = b
ans = [](0x0)

Please note the complete different answer !

> numel(foo)
ans =  4

> numel(foo(:))
ans =  4

> numel(foo{:})
ans = 0


> size(foo)
ans =

   2   2

> size(foo(:))
ans =

   4   1

> size(foo{:})
error: Invalid call to size.  Correct usage is:


I supect you are thinking as foo(:) and foo{:} as equivalent
but they are not.

Regards
Marco