Re: char behavior

[Ben Abbott]

On Oct 9, 2011, at 1:35 AM, Rik wrote:

> On 10/08/2011 08:35 PM, Ben Abbott wrote:
>> On Oct 8, 2011, at 11:04 PM, Rik <address@hidden> wrote:
>> 
>>> On 10/08/2011 06:47 PM, address@hidden wrote:
>>>> Running the developers sources (default branch), I see ...
>>>> 
>>>>   size (char (1, 0))
>>>>   ans =   2   1
>>>> 
>>>> Does anyone else see this?
>>>> 
>>>> Ben
>>> Yes.  This is what you expect.  char() takes its arguments and concatenates
>>> them vertically.  There were two input arguments so the resulting 2-D
>>> character array will have two rows.
>>> 
>>> It's easier to see with this code:
>>> 
>>> char (97, 98)
>>> ans =
>>> 
>>> a
>>> b
>>> 
>>> size (ans)
>>> ans =
>>> 
>>>  2   1
>>> 
>>> --Rik
>> There is an incompatibility problem. 
>> 
>> With char(M,N) ML returns a character array with size [MxN]
>> 
>> Thus, char() should behave like ones(), zeros(), cell(), etc.
>> 
>> Ben
> Is this more undocumented Matlab weirdness?  I don't see this behavior in
> their documentation (http://www.mathworks.com/help/techdoc/ref/char.html). 
> What character did they choose to fill the new character matrix with?
> 
> This code snippet should produce a 5x3 matrix and show the integer value of
> whatever fill character they used.
> 
> double (char (5, 3))
> 
> --Rik

Looking at this after again, it appears I was imagining things yesterday.

Octave is working as ML does. 

Ben