Try
z = g.location.x;
and if g is scalar
z = [ g.location.x ];
as you would do with structures
.
Added: 2012-12-14
I fail to find a concise description on how Matlab is supposed to behave in these cases (R2012a). However, I found some "tips" and "examples". I need something better.
The statement
z = [ g(:).location(:).x ];
or short
z = [ g.location.x ];
consists of two operations: list expansion and concatenation. I don't find the documentation on "list expansion". However, I'm positive that "list expansion" only works with the rightmost index in a train-wreck like g.location.x
Here is an experiment with list expansion on objects. I think this illustrates pretty much what is possible to achieve.
>> g34 = ArrayClass( 3, 4 );
>> g14 = ArrayClass( 1, 4 );
>> g41 = ArrayClass( 4, 1 );
where
classdef ArrayClass < handle
properties
location
end
methods
function this = ArrayClass( n, m )
if nargin == 0, return, end
this( 1, n ) = ArrayClass();
for ii = 1 : n
this( ii ).location = XYZ( m );
end
end
end
end
and where
classdef XYZ < handle
properties
x
y
z
end
methods
function this = XYZ( m )
if nargin == 0, return, end
this( 1, m ) = XYZ();
for ii = 1 : m
this( ii ).x = 10 + rand;
this( ii ).y = 20 + rand;
this( ii ).z = 30 + rand;
end
end
end
end
and some calls
>> [ g14.location.x ]
ans =
10.5211 10.6241 10.3674 10.8852
>> [ g34(3).location.x ]
ans =
10.0855 10.0292 10.4886 10.4588
>> [ g41.location.x ]
Dot name reference on non-scalar structure.
>> [ g41(2).location.x ]
ans =
10.6797
.
"[ g41.location.x ]" breaks the the rightmost index rule.
