/*
* org.openmicroscopy.shoola.util.mem.Handle
*
*------------------------------------------------------------------------------
* Copyright (C) 2006 University of Dundee. All rights reserved.
*
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License along
* with this program; if not, write to the Free Software Foundation, Inc.,
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
*
*------------------------------------------------------------------------------
*/
package org.openmicroscopy.shoola.util.mem;
//Java imports
//Third-party libraries
//Application-internal dependencies
/**
* Provides the basic machinery to share the same logical state across objects
* with different identities.
* <p>This class calls for a distinction between object identity and object
* state. Made this distinction, it becomes possible to share the same state
* across different objects upon copy operations. When an object is updated,
* a new state representation is bound to that object, while the other objects
* can still share the previous state. This way we can have shallow copy with
* the semantics of deep copy. This can result in dramatically reduced memory
* footprint when:</p>
* <ul>
* <li>A considerable amount of copies of a given master object are
* needed.</li>
* <li>You can't use references to the master object, because the copied
* objects need to have their own identity.</li>
* <li>The number of copied objects that are going to change their state
* after the copy operation is small compared to the total number of
* copies.</li>
* </ul>
* <p>For example, think of a class <code>R</code> that represents a rectangle
* in the plane with <code>4</code> integer fields <code>x, y, w, h</code>, and
* say you want to use instances of this class to describe an ROI (region of
* interest) selection in a given image 3D-stack composed of <code>100</code>
* planes — each plane would contain a rectangle selection and all those
* selections would make up your ROI. Let's assume that the initial selection
* is a discrete 3D-rectangle that spans all planes in the stack — you
* would have one rectangle per plane, every rectangle would have exactly the
* same state, say <code>s[x=0, y=0, w=3, h=4]</code>. Moreover, let's assume
* that you will have to modify slightly this initial ROI in order to get the
* final selection — for example by resizing/moving a couple of rectangles
* within the selection. Now, when you start off with the initial selection,
* you could decide to clone an initial master object whose state is
* <code>s</code> — this way, you can later modify one of the copies
* without affecting the others. However, because <i>Java</i> makes no
* distinction between object identity and state, you would have in memory
* <code>100</code> references and <code>100</code> copies of the same logical
* state <code>s</code>, while you actually only need one.<p>
* <p>The purpose of this class is to help you save memory in situations like
* that just described by approximating the semantics of the well known
* Handle/Body and Counted Body idioms often found in <i>C++</i> programs. A
* given class abstraction is implemented by two actual classes which replicate
* the same class interface. One class, the Handle, takes on the role of an
* object identifier and forwards all calls to the other class, the Body, which
* implements the actual functionality. Clients can only access instances of
* the Handle which can all share the same Body object whenever appropriate.</p>
* <p>The way this works in our case is pretty easy. An Handle class extends
* this base <code>Handle</code> class and provides a reference to an instance
* of the corresponding Body class. The concrete Handle class exposes the same
* interface as its corresponding Body (this is not an absolute requirement, but
* usually an implementation trade-off) and has <i>no state</i> — in fact,
* the state is hold by the associated Body object. The Handle just forwards to
* the Body any call that only reads the Body's state. However, it must call
* the {@link #breakSharing() breakSharing} protected method <i>before</i>
* forwarding any call that modifies the Body's state. It is crucial that
* concrete <code>Handle</code> classes stick to this rule. In fact, the
* {@link #copy() copy} method simply rebinds a new <code>Handle</code> to the
* existing Body, so subclasses must notify any incumbent change to the Body's
* state for the <code>Handle</code> to break state sharing. Lastly, it's also
* fundamental that the Body class implements the {@link Copiable} interface
* correctly for all this to work properly.</p>
*
* @author Jean-Marie Burel
* <a href="mailto:j.burel@dundee.ac.uk">j.burel@dundee.ac.uk</a>
* @author <br>Andrea Falconi
* <a href="mailto:a.falconi@dundee.ac.uk">
* a.falconi@dundee.ac.uk</a>
* @version 2.2
* <small>
* (<b>Internal version:</b> $Revision$ $Date$)
* </small>
* @since OME2.2
*/
public abstract class Handle
implements Copiable, Cloneable
{
/** Reference to the Body object. */
private Copiable body;
/**
* Tells whether the {@link #body} is referenced by other
* <code>Handle</code> objects.
*/
private boolean shared;
/**
* Subclasses use this constructor to specify the Body instance this
* handle will be paired up with.
* Subclasses must pass in a <i>newly</i> created object.
*
* @param body Reference to the Body object. Mustn't be <code>null</code>.
*/
protected Handle(Copiable body)
{
if (body == null)
throw new NullPointerException("No body.");
this.body = body;
shared = false;
}
/**
* Returns a reference to the Body object that is <i>currently</i> paired
* up with this handle.
* The type of the returned object is the same as the one of the object
* that was passed to this class' protected constructor. However, the
* object returned by this method could be different from the one initially
* passed in at creation time if the {@link #breakSharing() breakSharing}
* method has been invoked. For this reason, subclasses mustn't cache a
* reference to the object returned by this method.
* Moreover, subclasses must never leak out a reference to the returned
* Body object.
*
* @return The Body object.
*/
protected Object getBody() { return body; }
/**
* Subclasses must call this method <i>before</i> forwarding any call
* that modifies the Body's state.
*/
protected final void breakSharing()
{
if (shared) {
body = (Copiable) body.copy();
shared = false;
}
}
/**
* Returns a deep copy of this object.
* To be precise, this method returns an object that will behave like
* a deep copy, but has a negligible memory footprint until an attempt
* to change its state is made. Then the whole original state is restored
* in memory so that the state change operation can take place.
*
* @return A deep copy of this object. The class of the returned object
* is the same as the class of this object.
*/
public final Object copy()
{
Handle h;
//Make a shallow copy of this object. This is fine b/c subclasses
//are not supposed to hold any state, never mind references to other
//objects :)
try {
h = (Handle) clone(); //Class of h is this instance's class.
} catch (CloneNotSupportedException cnse) {
//Shouldn't happen as this class implements Cloneable.
throw new InternalError(
"JVM Internal Error: couldn't clone object that "+
"implements Cloneable.");
}
h.body = this.body; //Not actually needed, added for clarity.
//Set state sharing flag.
h.shared = true;
this.shared = true;
return h;
}
}