Ada Programming/Function overloading
Description
Function overloading (also method overloading) is a programming concept that allows programmers to define two or more functions with the same name and in the same scope.
Each function has a unique signature (or header), which is derived from:
- function/procedure name
- number of arguments
- arguments' type
- arguments' order
- arguments' name
- return type
Please note: Not all above signature options are available in all programming languages.
| Language | 1 | 2 | 3 | 4 | 5 | 6 |
|---|---|---|---|---|---|---|
| Ada | yes | yes | yes | yes | yes | yes |
| C++ | yes | yes | yes | yes | no | no |
| Java | yes | yes | yes | yes | no | no |
| Swift | yes | yes | yes | yes | yes | yes |
Warning: Function overloading is often confused with function overriding. In Function overloading, a function with a different signature is created, adding to the pool of available functions. In function overriding, however, a function with the same signature is declared, replacing the old function in the context of the new function.
Demonstration
Since functions' names are in this case the same, we must preserve uniqueness of signatures, by changing something from the parameter list (last three alienees).
If the functions' signatures are sufficiently different, the compiler can distinguish which function was intended to be used at each occurrence. This process of searching for the appropriate function is called function resolution and can be quite an intensive one, especially if there are a lot of equally named functions.
Programming languages supporting implicit type conventions usually use promotion of arguments (i.e. type casting of integer to floating-point) when there is no exact function match. The demotion of arguments is rarely used.
When two or more functions match the criteria in function resolution process, an ambiguity error is reported by compiler. Adding more information for the compiler by editing the source code (using for example type casting), can address such doubts.
The example code shows how function overloading can be used. As functions do practically the same thing, it makes sense to use function overloading.
functionGenerate_Number (MaxValue : Integer)returnIntegerissubtypeRandom_TypeisIntegerrange0 .. MaxValue;packageRandom_PackisnewAda.Numerics.Discrete_Random (Random_Type); G : Random_Pack.Generator;beginRandom_Pack.Reset (G);returnRandom_Pack.Random (G);endGenerate_Number;functionGenerate_Number (MinValue : Integer; MaxValue : Integer)returnIntegerissubtypeRandom_TypeisIntegerrangeMinValue .. MaxValue;packageRandom_PackisnewAda.Numerics.Discrete_Random (Random_Type); G : Random_Pack.Generator;beginRandom_Pack.Reset (G);returnRandom_Pack.Random (G);endGenerate_Number;
calling the first function
The first code block will generate numbers from 0 up to specified parameter MaxValue. The appropriate function call is:
Number_1 : Integer := Generate_Number (10);
calling the second function
The second requires another parameter MinValue. Function will return numbers above or equals MinValue and lower than MaxValue.
Number_2 : Integer := Generate_Number (6, 10);
Function overloading in Ada
Ada supports all six signature options, but if you use the arguments' name as option, you will always have to name the parameter when calling the function. i.e.:
Number_2 : Integer := Generate_Number (MinValue => 6,
MaxValue => 10);
Note that you cannot overload a generic procedure or generic function within the same package. The following example will fail to compile:
packagemyPackagegenerictypeValue_Typeis(<>); -- The first declaration of a generic subprogram -- with the name "Generic_Subprogram"procedureGeneric_Subprogram (Value :inoutValue_Type); ...generictypeValue_Typeis(<>); -- This subprogram has the same name, but no -- input or output parameters. A non-generic -- procedure would be overloaded here. -- Since this procedure is generic, overloading -- is not allowed and this package will not compile.procedureGeneric_Subprogram; ...generictypeValue_Typeis(<>); -- The same situation. -- Even though this is a function and not -- a procedure, generic overloading of -- the name "Generic_Subprogram" is not allowed.functionGeneric_Subprogram (Value : Value_Type)returnValue_Type;endmyPackage;