Skip to main content

Lab: Queues, arrays, and priority queues

In this laboratory, you will have an opportunity to ground your understanding of queues, particularly of the array-based implementation of queues. You will also continue to explore some other related topics.


a. Open the reading on linear structures, the reading on queues, the reading on priority queues, and the reading on wrapper classes in new tabs.

b. If you haven’t already done so, fork and clone the repo at


Exercise 1: Code reading

Read through You will note that the iterator is not yet implemented. That’s okay; we’ll talk about iterators in the near future. More importantly, you may also note a few subtle (or not so subtle) bugs. If you do, write them down. If not, that’s okay, we’ll work them out in the lab.

Exercise 2: Some basic experiments

Look at Take notes as to what the queue should look like at each step of the first series of procedure calls. You may also want to revisit the ReportingLinearStructure class to recall how it works.

Run ArrayBasedQueueExpt and see if you get the output that you expect.

Exercise 3: Squashing bugs

You’ve probably determined that there seem to be some significant bugs in the queue implementation. Can you tell where they are?

You might learn a bit more about the bug by adding a call to expt.put("d") before the first call to expt.get().

Do your best to correct the first bug: get and peek don’t seem to return the correct value after some point. If you need a hint as to where to look, ask your instructor or mentor.

Exercise 4: Wrap-around in arrays

If you uncomment the second section of code and reduce the size of the queue to, say, 4, you may find that the queue fills before it should. (You may also have dealt with that issue.)

How do we fix that problem? Normally, we “wrap around”, so that the back of the queue goes to the front of the array. For example, if we have seven items in a queue, and the front is at 4, then the item 0 is at 4; item 1 is at 5; item 2 is at 6; item 3 is at 7; item 4 can’t be at 8 (there is no index 8), so we wrap it around to index 0; item 5 is at 1, item 6 is at 2, and the back of the queue is at 3.

Rewrite your code so that elements wrap in the specified way. You’ll need to change back. You may also need to change the code for isEmpty and isFull.

Exercise 5: Extending the reporter

The reading on wrapper classes suggested that we could make a one-parameter constructor for something liked ReportingLinkedStructure that (a) sets pen to a PrintWriter that prints to stderr, and (b) sets name to the class name of the wrapped class and what appears to be a useful identifier.

Add that code and verify that it works as advertised. If not, figure out how to correct it.

Exercise 6: Testing

Up to now, we’ve been exploring our linear structures by manually comparing actual output to expected output. As we’ve learned, computers are much better than humans at identifying trouble spots.

a. Read through the PriorityQueue interface we’ve provided to refresh your understanding of how priority queues are supposed to behave.

b. Unfortunately (or perhaps fortunately), it is difficult to write a test suite for an interface. So you will instead write a test suite for BuiltinPriorityQueue, which is similar to the JUPQadapter from the reading on wrapper classes.

Your test suite should be sufficiently sophisticated that you can be relatively confident that it will catch appropriate errors.

The PriorityQueue interface tells you how they should behave, so the only additional information you need is the constructor, which has the following form.

   * Create a new priority queue that holds up to capacity elements and 
   * uses order to compare elements.
  public BuiltinPriorityQueue(int capacity, Comparator<T> order) throws Exception {
    // ...
  } // BuiltinPriorityQueue(int capacity, Comparator<T>)

Here are a few comparators you may find useful. (In the future, we’ll learn how to write these more concisely using lambdas or anonymous inner classes.)

class StringComparator implements Comparator<String> {
  public int compare(String str1, String str2) {
    // Efficiency hack: If two strings occupy the same memory
    // they are equal.
    if (str1 == str2) { return 0; }
    // Safety check: If either string is null, compareTo may fail,
    // so we make sure neither is null.  We treat null as "smaller"
    // than any other string.
    if (str1 == null) { return -1; }
    if (str2 == null) { return 1; }
    // Finally, we can use the built-in `compareTo` method.
    return str1.compareTo(str2);
  } // compare(String, STring)
} // StringComparator

class IntComparator implement Comparator<Integer> {
  public int compare(Integer i, Integer j) {
    // While this method sometimes gets implemented as i-j, that
    // implementation presents overflow risks, so we choose a
    // somewhat more verbose approach.
    if (i < j) { return -1; }
    else if (j < i) { return 1; }
    else return 0;
  } // compare(Integer, Integer)
} // IntegerComparator

Exercise 7: Adapting classes

a. Remind yourself of the methods specified by our LinearStructure interface.

b. Skim through the documentation on the standard Java implementation of priority queues, available at

c. Discuss with your partner how you would write an adapter class to make the built in priority queues match the desired behavior of our priority queue interface.

d. Look at to see how our implementation matches your design.

e. Run your test suite.

For those with extra time

If you are fortunate enough to have some extra time, you might consider doing any of the following.

Extra 1: Linked queues

If you did not finish implementing linked queues in the previous lab, do so now.

Extra 2: Array-based priority queues

a. Read through You will note that the iterator is not yet implemented and that prioritization is not yet implemented.

b. Make some notes to yourself as to how you might finish implementing the put and get methods.

c. As the reading noted, there are two basic strategies for implementing priority queues in arrays.

  • You can keep the values in order from lowest priority to highest priority. In this case, the put method must ensure that the elements in the array are stored in order. (You can probably use a variant of the insert method from insertion sort to achieve that goal.)
  • You can keep the values in arbitrary order and search for the highest-priorty element whenever we call get or peek.

Pick one and finish the implementation.


Execises 1-4 are taken primarily from a lab on array-based queues from CSC 207 2014F

Exercise 7 and Extra 2 are taken primarily from a lab on priority queues from CSC 207 2014F

The remaining exercises are new to this lab.

All of these materials were written by Samuel A. Rebelsky.