{"id":95,"date":"2006-11-21T07:00:52","date_gmt":"2006-11-21T12:00:52","guid":{"rendered":"https:\/\/blogs.mathworks.com\/steve\/?p=95"},"modified":"2019-10-22T16:33:24","modified_gmt":"2019-10-22T20:33:24","slug":"old-habits","status":"publish","type":"post","link":"https:\/\/blogs.mathworks.com\/steve\/2006\/11\/21\/old-habits\/","title":{"rendered":"Old habits"},"content":{"rendered":"
\r\n

Last time<\/a> I showed off custom data cursors with a little utility function that displays object labels as data cursors:\r\n <\/p>\r\n

<\/p>\r\n

My code used label2rgb<\/tt> to display a \"colorized\" label matrix. The custom data cursor callback, though, needed the original label matrix, so I used\r\n setappdata<\/tt> to save the label matrix in a place where the callback routine could retrieve it.\r\n <\/p>\r\n

Within a few hours of posting that code, I had a \"Doh!\" moment. I haven't used that kind of setappdata<\/tt> trick since nested function handles appeared in MATLAB 7.0. I have no idea why I suddenly reverted to my previous coding\r\n patterns. Old habits die hard, as the cliche says.\r\n <\/p>\r\n

Anyway, nested function handles in MATLAB 7 can be used to create callback functions that carry around their own state. I\r\n thought I would show you how my display_label_matrix<\/tt><\/a> function looks using this technique.\r\n <\/p>

type display_label_matrix<\/span><\/pre>
\r\nfunction display_label_matrix(L)\r\n% display_label_matrix Display label matrix with custom data cursors\r\n% display_label_matrix(L) displays the label matrix L as colored objects on\r\n% a gray background.  It also installs a custom data cursor handler in the\r\n% figure.  In data cursor mode, clicking on an object displays a data\r\n% cursor containing the object's number.\r\n\r\n% Steve Eddins\r\n% Copyright 2006 The MathWorks, Inc.\r\n% $Revision: 1.2 $  $Date: 2006\/11\/21 02:41:00 $\r\n\r\nrgb = label2rgb(L, 'jet', [.7 .7 .7], 'shuffle');\r\nh = imshow(rgb);\r\n\r\n% Enable the figure data cursor mode, and use our own custom data cursor\r\n% strings.\r\ndcm = datacursormode(ancestor(h, 'figure'));\r\nset(dcm,'Enable','on', ...\r\n    'Updatefcn', @dataCursorText);\r\n\r\n    %======================================================================\r\n    function output_text = dataCursorText(obj, event_obj)\r\n    % Respond to a user click in data cursor mode by displaying the label\r\n    % number of the pixel that was clicked on.\r\n\r\n        h_image = get(event_obj, 'Target');\r\n        pos = get(event_obj, 'Position');\r\n        clicked_label = L(round(pos(2)), round(pos(1)));\r\n        output_text = sprintf('%d', clicked_label);\r\n    end\r\n    %----------------------------------------------------------------------\r\n    \r\nend\r\n\r\n\r\n\r\n\r\n<\/pre>
In the new version, the callback function, dataCursorText<\/tt>, is nested within the outer function, display_label_matrix<\/tt>.  Notice how dataCursorText<\/tt> uses a variable, L<\/tt>, that is created in the scope of the outer function, display_label_matrix<\/tt>.  When the outer function creates a function handle to dataCursorText<\/tt>, the value of L<\/tt> at that moment is captured into a kind of private workspace that goes along with the function handle.  No auxiliary information\r\n      or state needs to be stored anywhere else; the function handle has inside of it all the information it needs to perform its\r\n      computation.\r\n   <\/p>\r\n   
If you are interested in learning more about nested functions in MATLAB, you might want to look at Loren Shure's News & Notes article<\/a> on the topic.\r\n   <\/p>