Posts Tagged ‘Memory’

JIDE Property Grids

Wednesday, April 21st, 2010

I would like to welcome guest blogger Levente Hunyadi.

Matlab’s property inspector

We often wish to edit properties of heterogeneous objects using a common interface. Matlab’s property inspector, invoked with the built-in inspect function, answers this need. The inspector is based on a two-column table of property names and values. Properties and their values are populated automatically, and the user can edit values in-place. The inspector enables property categorization, sub-categorization and sorting, which help users find and modify properties easily. For each property, the inspector displays a matching edit control: editbox/combobox/checkbox etc. This simplifies property value editing and prevents illegal value entry. Matlab’s GUI builder, GUIDE, uses the inspector to let users edit GUI properties such as position, color etc. It is also used by other tools such as the Plot Editor.

Matlab's built-in property inspector

Matlab's built-in property inspector

The Matlab inspector can be embedded, with not-too-much effort, within Matlab GUI applications. Examples of this can be found in the FindJObj and UIInspect utilities.

FindJObj - embedded property inspector

FindJObj - embedded property inspector

Unfortunately, Matlab’s property inspector is limited to Handle Graphics, Java and COM objects, and cannot be used for structures or user-defined Matlab classes. We shall see below how to set up our own property grid, populate it with data, and subscribe to property change events. This is a rather procedural approach. It is usually more convenient to use a declarative approach in which a structure or Matlab class is passed to a function that automatically discovers its properties and their meta-information. The Property Grid utility at Matlab File Exchange provides these services.

A simple property grid

Matlab’s property inspector is based on a property grid control by JIDE Software. JIDE Grids is a collection of components that extend the standard Java Swing JTable component, and is included in each Matlab installation (/java/jarext/jide/jide-grids.jar under the Matlab root). In particular, JIDE Grids includes the PropertyTable class, which is a fully customizable property grid component. You can find further details on JIDE Grids in the Developer Guide and the Javadoc documentation.

There are several related classes associated with the PropertyTable class. First, a PropertyTableModel encapsulates all properties that are visualized in the property grid. Each property derives from the Property abstract class, which features some common actions to properties, most notably to get and set property value. DefaultProperty is a default concrete subclass of Property. Finally, PropertyPane decorates a property grid with icons for changing category view to alphabetically sorted view as well as expanding and collapsing categories, and a description text box at the bottom that can be shown or hidden.

Here are the DefaultProperty fields and their respective roles:

Field Role
Name Interal property name, not necessarily displayed, used as a key to identify the property.
DisplayName A short property name shown in the left column of the property grid.
Description A concise description of the property, shown at the bottom of the property pane, below the grid.
Type The Java type associated with the property, used to invoke the appropriate renderer or editor.
EditorContext An editor context object. If set, both the type and the context are used to look up the renderer or editor to use. This lets, for instance, one flag value to display as a true/false label, while another as a checkbox.
Category A string specifying the property’s category, for grouping purposes.
Editable Specifies whether the property value is modifiable or read-only.
Value The current property value, as a Java object.

Just as with any Java object, these fields may either be accessed with the Java get/set semantics (e.g. getName() or setName(name)), or the Matlab get/set semantics (e.g. get(prop,’Name’) or set(prop,’Name’,name)). When using the Matlab syntax, remember to wrap the Java object in a handle() call, to prevent a memory leak.

To use a property grid in Matlab, first construct a set of DefaultProperty objects. For each object, set at least the name, type and initial value. Next, add the properties to a table model. Finally, construct a property grid with the given table model and encapsulate in a property pane:

% Initialize JIDE's usage within Matlab
com.mathworks.mwswing.MJUtilities.initJIDE;
 
% Prepare the properties list
list = java.util.ArrayList();
prop1 = com.jidesoft.grid.DefaultProperty();
prop1.setName('stringProp');
prop1.setType(javaclass('char',1));
prop1.setValue('initial value');
prop1.setCategory('My Category');
prop1.setDisplayName('Editable string property');
prop1.setDescription('A concise description for my property.');
prop1.setEditable(true);
list.add(prop1);
 
prop2 = com.jidesoft.grid.DefaultProperty();
prop2.setName('flagProp');
prop2.setType(javaclass('logical'));
prop2.setValue(true);
prop2.setCategory('My Category');
prop2.setDisplayName('Read-only flag property');
prop2.setEditable(false);
list.add(prop2);
 
% Prepare a properties table containing the list
model = com.jidesoft.grid.PropertyTableModel(list);
model.expandAll();
grid = com.jidesoft.grid.PropertyTable(model);
pane = com.jidesoft.grid.PropertyPane(grid);
 
% Display the properties pane onscreen
hFig = figure;
panel = uipanel(hFig);
javacomponent(pane, [0 0 200 200], panel);
 
% Wait for figure window to close & display the prop value
uiwait(hFig);
disp(prop1.getValue());

Here, com.mathworks.mwswing.MJUtilities.initJIDE is called to initialize JIDE’s usage within Matlab. Without this call, we may see a JIDE warning message in some Matlab releases. We only need to initJIDE once per Matlab session, although there is no harm in repeated calls.

javaclass is a function (included in the Property Grid utility, or directly downloadable from here) that returns a Java class for the corresponding Matlab type with the given dimension: javaclass(‘logical’) or javaclass(‘logical’,0) (a single logical flag value) returns a java.lang.Boolean class; javaclass(‘char’,1) (a character array) returns a java.lang.String class; javaclass(‘double’,2) (a matrix of double-precision floating point values) returns double[][].

javacomponent is the undocumented built-in Matlab function that adds Java Swing components to a Matlab figure, using the given dimensions and parent handle. When the user closes the figure, prop.getValue() fetches and displays the new property value.

A simple user-defined property grid

A simple user-defined property grid

Next week’s article will show how to add more complex renderers and editors (display the flag value as a checkbox for example), define nested properties, and subscribe to property value change events. So stay tuned…

Undocumented profiler options

Thursday, April 2nd, 2009

The Matlab profiler is a powerful tool for debugging performance-related issues in Matlab applications. However, it also has some undocumented options that facilitate other forms of debugging, namely memory bottlenecks and JIT (Just-In-Time compilation) problems.

To turn on memory stats in the profile report, run this (only once is necessary – will be remembered for future profiling runs):

profile -memory on;
profile('-memory','on');  % an alternative

To turn on JIT information, run this (again, only once is necessary, prior to profile report):

setpref('profiler','showJitLines',1);

You will then see additional JIT and memory (allocated, freed and peak) information displayed in the profile report, as well as the options to sort by allocated, freed and peak memory:


Profile report with memory & JIT infoProfile report with memory & JIT info

Profile report with memory & JIT info

Profile report with memory & JIT info

For those interested, the references to these two options appear within the code of profview.m (line 1199 on R2007b), for the JIT option:

showJitLines = getpref('profiler','showJitLines',false);

…and profile.m (lines 163-165 on R2007b), for the memory option:

if memory ~= -1
    callstats('memory', memory);
end

Note that there appears to be two undocumented additional memory-related options in profile.m (lines 311-312):

options = {'detail', 'timer', 'history', 'nohistory', 'historysize', ...
           'timestamp', 'memory', 'callmemory', 'nomemory' };

However, ‘-nomemory’ appears to simply turn the memory stats collection off, and ‘-callmemory’ is not recognized because of a bug in line 349, which looks for ‘callnomemory’…:

    case 'callnomemory'   % should be 'callmemory'
           memory = 2;

When this bug is fixed, we see that we get only partial memory information, so the ‘-callmemory’ option is really not useful – use ‘-memory’ instead.