Objects that serve as keys in ordered sets and maps should be immutable. When some fields must be mutable, the equals()
, hashCode()
and compareTo()
methods must consider only immutable state when comparing objects. Violations of this rule can produce inconsistent orderings in collections. The documentation of java.util.Interface Set<E>
and java.util.Interface Map<K,V>
warns against this. For example, the documentation for the Interface Map states [[API 2006]]:
Note: great care must be exercised [when] mutable objects are used as map keys. The behavior of a map is not specified if the value of an object is changed in a manner that affects
equals
comparisons while the object is a key in the map. A special case of this prohibition is that it is not permissible for a map to contain itself as a key. While it is permissible for a map to contain itself as a value, extreme caution is advised: theequals
andhashCode
methods are no longer well defined on such a map.
Noncompliant Code Example
This noncompliant code example defines a mutable class Employee
that consists of the fields name
and salary
, whose values can be changed using the respective setters. The equals()
method is overridden to provide a comparison facility by employee name.
// Mutable class Employee class Employee { private String name; private double salary; Employee(String empName, double empSalary) { this.name = empName; this.salary = empSalary; } public void setEmployeeName(String empName) { this.name = empName; } // ... including a hashCode implementation @Override public boolean equals(Object o) { if (!(o instanceof Employee)) { return false; } Employee emp = (Employee)o; return emp.name.equals(name); } } // Client code Map<Employee, Calendar> map = new ConcurrentHashMap<Employee, Calendar>(); // ...
Use of the Employee
object as a key to the map is insecure because the properties of the object may change after an ordering has been established. For example, a client may modify the name
field when the last name of an employee changes. As a result, clients may observe non-deterministic behavior.
Compliant Solution
This compliant solution adds a final field employeeID
that is immutable after initialization. The equals()
method compares Employee
objects on the basis of this field.
// Mutable class Employee class Employee { private String name; private double salary; private final long employeeID; // Unique for each Employee Employee(String name, double salary, long empID) { this.name = name; this.salary = salary; this.employeeID = empID; } // ... including a hashCode implementation @Override public boolean equals(Object o) { if (!(o instanceof Employee)) { return false; } Employee emp = (Employee)o; return emp.employeeID == employeeID; } } // Client code remains same Map<Employee, Calendar> map = new ConcurrentHashMap<Employee, Calendar>(); // ...
The Employee
class can now be safely used as a key for the map in the client code.
Noncompliant Code Example
Many programmers are surprised by an instance of hash code mutability that arises because of serialization. The contract for the hashCode()
method lacks any requirement that hash codes remain consistent across different executions of an application. Similarly, when an object is serialized and subsequently deserialized, its hashcode after deserialization may be inconsistent with its original hashcode.
This noncompliant code example uses the Key
class as the key index for the Hashtable
. The Key
class may or may not override Object.equals()
, but it distinctly does not override Object.hashCode()
. According to the Java API [[API 2006]] class Hashtable
documentation,
To successfully store and retrieve objects from a hash table, the objects used as keys must implement the
hashCode
method and theequals
method.
This noncompliant code example follows the above advice but can nevertheless fail after serialization and deserialization. Consequently, it may be impossible to retrieve the value of the object after deserialization by using the original key.
class Key implements Serializable { // Does not override hashCode() } class HashSer { public static void main(String[] args) throws IOException, ClassNotFoundException { Hashtable<Key,String> ht = new Hashtable<Key, String>(); Key key = new Key(); ht.put(key, "Value"); System.out.println("Entry: " + ht.get(key)); // Retrieve using the key, works // Serialize the Hashtable object FileOutputStream fos = new FileOutputStream("hashdata.ser"); ObjectOutputStream oos = new ObjectOutputStream(fos); oos.writeObject(ht); oos.close(); // Deserialize the Hashtable object FileInputStream fis = new FileInputStream("hashdata.ser"); ObjectInputStream ois = new ObjectInputStream(fis); Hashtable<Key, String> ht_in = (Hashtable<Key, String>)(ois.readObject()); ois.close(); if (ht_in.contains("Value")) // Check whether the object actually exists in the hash table System.out.println("Value was found in deserialized object."); if (ht_in.get(key) == null) // Gets printed System.out.println("Object was not found when retrieved using the key."); } }
Compliant Solution
This compliant solution changes the type of the key value to be an Integer
object. Consequently, key values remain consistent across multiple runs of the program, across serialization and deserialization, and also across multiple JVMs.
class HashSer { public static void main(String[] args) throws IOException, ClassNotFoundException { Hashtable<Integer, String> ht = new Hashtable<Integer, String>(); ht.put(new Integer(1), "Value"); System.out.println("Entry: " + ht.get(1)); // Retrieve using the key // Serialize the Hashtable object FileOutputStream fos = new FileOutputStream("hashdata.ser"); ObjectOutputStream oos = new ObjectOutputStream(fos); oos.writeObject(ht); oos.close(); // Deserialize the Hashtable object FileInputStream fis = new FileInputStream("hashdata.ser"); ObjectInputStream ois = new ObjectInputStream(fis); Hashtable<Integer, String> ht_in = (Hashtable<Integer, String>)(ois.readObject()); ois.close(); if (ht_in.contains("Value")) // Check whether the object actually exists in the Hashtable System.out.println("Value was found in deserialized object."); if (ht_in.get(1) == null) // Not printed System.out.println("Object was not found when retrieved using the key."); } }
This problem can also be avoided by overriding the the hashcode()
method in the Key
class, though it is best to avoid serializing hash tables that are known to use implementation-defined parameters.
Risk Assessment
Failure to ensure that the keys used in a comparison operation are immutable can lead to non-deterministic behavior.
Rule |
Severity |
Likelihood |
Remediation Cost |
Priority |
Level |
---|---|---|---|---|---|
MET11-J |
low |
probable |
high |
P2 |
L3 |
Automated Detection
The Coverity Prevent Version 5.0 MUTABLE_COMPARISON checker can detect the instances where compareTo method is reading from a non-constant field. If the non-constant field is modified, the value of compareTo might change, which may break program invariants.
Bibliography
<ac:structured-macro ac:name="unmigrated-wiki-markup" ac:schema-version="1" ac:macro-id="f1bfe5f9-d7ac-4a78-bb47-e7cc883aa19e"><ac:plain-text-body><![CDATA[ |
[[API 2006 |
AA. Bibliography#API 06]] |
|
]]></ac:plain-text-body></ac:structured-macro> |