Question

In my custom view I’m looking into using Canvas.getClipBounds() to optimize my onDraw method (so that I’m only drawing what’s absolutely necessary each time it’s called).

However, I still want to absolutely avoid any object creation…

My question, therefore is: does getClipBounds() allocate a new Rect each time it is called? Or is it simply recycling a single Rect?

And if it is allocating a new object, can I save this expense by using getClipBounds(Rect bounds), which seems to use the passed Rect instead of its own?

---

(before you scream premature optimization, do consider that when placed in a ScrollView, onDraw can be called many times each second)

Answer 1

I have looked into the source code for Canvas as of Android 4.0.1, and it is as follows:

/**
 * Retrieve the clip bounds, returning true if they are non-empty.
 *
 * @param bounds Return the clip bounds here. If it is null, ignore it but
 *               still return true if the current clip is non-empty.
 * @return true if the current clip is non-empty.
 */
public boolean getClipBounds(Rect bounds) {
    return native_getClipBounds(mNativeCanvas, bounds);
}


/**
 * Retrieve the clip bounds.
 *
 * @return the clip bounds, or [0, 0, 0, 0] if the clip is empty.
 */
public final Rect getClipBounds() {
    Rect r = new Rect();
    getClipBounds(r);
    return r;
}

So, answering your question, getClipBounds(Rect bounds) will spare you from one object creation, but getClipBounds() will actually create a new Rect() object every time it is called.