Re: arrays of objects and builtin methods

[Ben Abbott]

On Dec 19, 2011, at 10:59 AM, J.J.Green wrote:

> Hi all
> 
> Consider the following trivial class
> 
> @myobj/myobj.m
> 
> function f = myobj(str)
>  f.name = str;
>  f = class(f,'myobj');
> 
> @myobj/display.m
> 
> function display(f)
>  fprintf('%s\n',f.name);
> 
> @myobj/transpose.m 
> 
> function f = transpose(g)
>  % do interesting things with input ...
>  f = builtin('transpose',g);
> 
> Now, lets make an array of these ...
> 
> GNU Octave, version 3.5.90+
> Copyright (C) 2011 John W. Eaton and others.
> :
> 
> octave:1> p = myobj('foo')
> foo
> octave:2> q = myobj('bar')
> bar
> octave:3> a = [p q]
> foo
> bar
> octave:4> size(a)
> ans =
> 
>   1   2
> 
> octave:5> class(a)
> ans = myobj
> octave:6> b = a.'
> error:   ../@myobj/transpose.m at line 3, column 3
> error:   ../@myobj/transpose.m at line 3, column 3
> :
> (and lots more of these)
> 
> It seems that the builtin array transpose is not "seen"
> in the overloaded transpose method for the object, so it
> calls itself again and again until some kind of recursion
> depth is exceeded.
> 
> The same code runs fine on Matlab
> 
> 
>                           < M A T L A B (R) >
>                  Copyright 1984-2011 The MathWorks, Inc.
>                    R2011b (7.13.0.564) 64-bit (glnxa64)
>                              August 13, 2011
> 
>>> p = myobj('foo')              
> foo
>>> q = myobj('bar')
> bar
>>> a = [p q]
> foo
> bar
>>> size(a)
> 
> ans =
> 
>     1     2
> 
>>> b = a.'
> foo
> bar
>>> size(b)
> 
> ans =
> 
>     2     1
> 
> Similar problems arise with other array methods (eg vertcat)
> 
> Any ideas?  Possible workarounds?
> 
> Thanks in advance
> 
> Jim

Looks like a bug.

Please file a report in the tacker.

        https://savannah.gnu.org/bugs/?group=octave

Ben