Constant interface

In the Java programming language, the constant interface pattern describes the use of an interface solely to define constants, and having classes implement that interface in order to achieve convenient syntactic access to those constants. However, since constants are very often merely an implementation detail, and the interfaces implemented by a class are part of its exported API, this practice amounts to putting implementations details into the API, which is considered inappropriate. ^[1][2] In general, collecting system constants into classes independent of behaviour, might create a poor object-oriented design, because it is often a sign of low cohesion. It is for these reasons that implementing constants interfaces is considered to be an anti-pattern.

Use of this anti-pattern has a few other downsides:

1. It pollutes the class namespace with read-only variables that may not be of use.
2. Contrary to the compile-time tactical utility of implementing a constants interface, the incidental run-time artifacts have little practical purpose (cf. marker interfaces which also have no methods but are useful at run-time).
3. If binary code compatibility is required in future releases, the constants interface must remain forever an interface (it cannot be converted into a class), even though it has not been used as an interface in the conventional sense.
4. Without an IDE that resolves where the constant are coming from, tracking it back to its containing class or interface can be time consuming.
5. A variable (representing an instance) of the interface is syntactically no more useful than the interface name itself (since it has no methods).

Example

public interface Constants {
 
	public static final double PI = 3.14159;
	public static final double PLANCK_CONSTANT = 6.62606896e-34;
}
 
 
public class Calculations implements Constants {
 
	public double getReducedPlanckConstant() {
		return PLANCK_CONSTANT / (2 * PI);
	}
}

Alternatives

Many of the pitfalls of the anti-pattern can be avoided by converting the constants interface to a proper class with no instances:

public final class Constants {
 
	private Constants() {
		// restrict instantiation
	}
 
	public static final double PI = 3.14159;
	public static final double PLANCK_CONSTANT = 6.62606896e-34;
}

This still leaves the original intent of the pattern mostly un-addressed (i.e. there is no syntax for accessing the constants unqualified). However, since Java 5, consider using static import^[2] to achieve the same goal:

import static Constants.PLANCK_CONSTANT;
import static Constants.PI;
 
public class Calculations {
 
	public double getReducedPlanckConstant() {
		return PLANCK_CONSTANT / (2 * PI);
	}
}

To varying degrees, the issues listed above have now been addressed:

1. Because static members can be imported specifically, the class namespace need not be polluted with all members of the constants interface.
2. Run-time and compile-time semantics are more closely aligned when using static imports instead of constants interfaces.
3. The compiled code has one less binary compatibility constraint (that "class Calculations implements Constants").
4. Because static imports apply only to the current file (and not the whole class hierarchy) it is easier to discover where each static member is declared.
5. There is less need to declare variables of the constants interface type, and it is potentially clearer that no concrete instances actually exist.

Note however, the changes do nothing to improve the cohesion of the Constants class, so static imports should not be considered to be a panacea.

References