SubRoutines

A subroutine is like a sub-program. It is not only helps divide your code up into sensible, manageable chunks, but it also allows these chunks to be used (called) by different parts of your program. Each subroutine contains one of more statements.

Functions are the same as procedures except that they return a value in addition to executing statements. A Function, as its name suggests, is like a little program that calculates something, returning the value to the caller. On the other hand, a procedure is like a little routine that performs something, and then just finishes.

Both functions and procedures can be defined to operate without any data being passed. For example, you might have a function that simply returns a random number (like the Pas2JS RandomInt function). It needs no data to get it going.

Likewise, you can have a procedure that carries out some task without the need for data to dictate its operations. For example, you might have a procedure that draws a square on the screen. The same square every time it is called.

Often, however, you will pass data, called parameters, to a subroutine. (Note that the definition of a subroutine refers to parameters as arguments - they are parameters when passed to the subroutine).

The following code illustrates simple function and procedure definitions:
Code example: procedure without parameters

procedure ShowTime; // A procedure with no parameters

begin

// Display the current date and time

WriteLn('Date and time is '+DateTimeToStr(Now));

end;

// Let us call this procedure

procedure TForm1.btnClearClick(Sender: TObject);

begin

ShowTime;

end;

procedure ShowTime(dateTime : Float); // With parameters

begin

// Display the date passed to the routine

WriteLn('Date and time is '+DateToStr(dateTime));

end;

function Today: Float;

begin

Result := Date;

end;

function Yesterday: Float;

begin

Result := Date - 1;

end;

function Tomorrow: Float;

begin

Result := Date + 1;

end;

// Let us call this procedure

procedure TForm1.btnClearClick(Sender: TObject);

begin

ShowTime(Today);

ShowTime(Yesterday);

ShowTime(Tomorrow);

end;

function RandomRange(const AFrom, ATo: Integer): Integer;

begin

if AFrom > ATo then

Result := RandomInt(AFrom - ATo) + ATo

else

Result := RandomInt(ATo - AFrom) + AFrom;

end;

function RandomChar : String;

var

i : integer;

begin

// Get a random number from 65 to 90

// (These numbers equate to characters 'A' to 'Z'

i := RandomRange(65, 90);

// Return this value as a char type in the return variable, Result

Result := Chr(i);

end;

procedure TForm1.btnClearClick(Sender: TObject);

begin

// Let us call this function

ShowMessage('Char chosen is : '+RandomChar);

end;

function RandomChar() {

var Result = "";

var i = 0;

i = RandomRange(65,90);

Result = Chr(i);

return Result

};

function RandomRange(AFrom, ATo) {

var Result = 0;

if (AFrom>ATo) {

Result = RandomInt(AFrom-ATo)+ATo;

} else {

Result = RandomInt(ATo-AFrom)+AFrom;

}

return Result

};

It is important to note that we return the value from a function in a special variable called Result that Pas2JS

secretly defines for us to be the same type as the return type of the function. We can assign to it at any point in the function.

When the function ends, the value then held in Result is then returned to the caller.

function Average(a, b, c : Extended) : Extended;

begin

// return the average of the 3 passed numbers

Result := Mean(a, b, c);

end;

// Let us call this function

ShowMessageFmt('Average of 2, 13 and 56 = %f',[Average(2,13,56)]);

NOTE: I would like to calculate the average of a set of three numbers. In Delphi you could use the mean function.

The Mean function returns the average of a set of Double values in a DataArray. As you can see, such function uses a

function SUM(const Data: array of Double): Extended;

asm // IN: EAX = ptr to Data, EDX = High(Data) = Count - 1

// Uses 4 accumulators to minimize read-after-write delays and loop overhead

// 5 clocks per loop, 4 items per loop = 1.2 clocks per item

FLDZ

MOV ECX, EDX

FLD ST(0)

AND EDX, not 3

FLD ST(0)

AND ECX, 3

FLD ST(0)

SHL EDX, 3 // count * sizeof(Double) = count * 8

JMP @Vector.Pointer[ECX*4]

@Vector:

DD @@1

DD @@2

DD @@3

DD @@4

@@4: FADD qword ptr [EAX+EDX+24] // 1

FXCH ST(3) // 0

@@3: FADD qword ptr [EAX+EDX+16] // 1

FXCH ST(2) // 0

@@2: FADD qword ptr [EAX+EDX+8] // 1

FXCH ST(1) // 0

@@1: FADD qword ptr [EAX+EDX] // 1

FXCH ST(2) // 0

SUB EDX, 32

JNS @@4

FADDP ST(3),ST // ST(3) := ST + ST(3); Pop ST

FADD // ST(1) := ST + ST(1); Pop ST

FWAIT

end;

But under Pas2JS, Assembler is pure fun – because to a browser assembler is JavaScript!

Example code : Calculate the mean of a set of 3 numbers using Smart Mobile Studio

function Mean(const Data: array of Float): Float;

var thisAverage : Variant;

begin

//Result := SUM(Data) / (High(Data) - Low(Data) + 1);

asm

var numberArray=[2,13,56], thisTotal=0,thisAverage=0;

// add elements of array together

for(var i=0;i<numberArray.length;i++)

{thisTotal+=numberArray[i];}

// calculate average

@thisAverage=(thisTotal/numberArray.length);

end;

// display result

Result := thisAverage;

end;

function Average(a, b, c : Float) : Float;

begin

// return the average of the 3 passed numbers

Result := Mean([a, b, c]);

end;

procedure TForm1.btnClearClick(Sender: TObject);

begin

// Let us call this function

ShowMessage(Format('Average of 2, 13 and 56 = %f',[Average(2,13,56)]));

end;

function Mean(Data) {

var Result = 0;

var thisAverage = undefined;

var numberArray=[2,13,56], thisTotal=0,thisAverage=0;

// add elements of array together

for(var i=0;i<numberArray.length;i++)

{thisTotal+=numberArray[i];}

// calculate average

thisAverage=(thisTotal/numberArray.length);

Result = Number(thisAverage);

return Result

};

function Average$1(a$56, b$18, c) {

return Mean(([a$56,b$18,c].slice()));

};

In our example above, we declared an integer variable for use in a calculation by the function.

Subroutines can have their own types, constants and variables, and these remain local to the routine.

Variable values are reset every time the routine is called (use a class object to hold onto data across routine calls).

procedure DoIt(A : Integer);

begin

A := A * 2;

ShowMessage(Format('A in the procedure = %d',[A]));

end;

procedure TForm1.InitializeForm;

var

A : Integer;

begin

inherited;

// this is a good place to initialize components

A := 22;

ShowMessage(Format('A in program before call = %d',[A]));

// Call the procedure

DoIt(A);

ShowMessage(Format('A in program now = %d',[A]));

end;

------------------------------
The procedure is passed A, updates it and displays it. The caller then displays the A that it passed to the procedure.

It is unchanged. The procedure sees this A as if it were defined as a local variable.

The default was of passing data is by what is called by value. Literally, the parameter value is passed to the subroutine argument.

reference to the argument is then to this copy of the variable value.
Passing by reference means that the subroutine actually refers to the passed variable rather than its value.

Any changes to the value will affect the caller variable. We declare a variable to be passed by reference with the var prefix.

procedure DoIt(var A : Integer);

begin

A := A * 2;

ShowMessage(Format('A in the procedure = %d',[A]));

end;

procedure TForm1.InitializeForm;

var

A : Integer;

begin

inherited;

// this is a good place to initialize components

A := 22;

ShowMessage(Format('A in program before call = %d',[A]));

// Call the procedure

DoIt(A);

ShowMessage(Format('A in program now = %d',[A]));

end;

Now the caller A variable is updated by the procedure.
This is a very useful way of returning data from a procedure. It also allows us to return more than one value from a subroutine.

For code clarity, and performance, it is often wise to declare arguments that are only ever read by a subroutine as constants. This is done with the const prefix. It can be used even when a non-constant parameter is passed. It simply means that the parameter is only ever read by the subroutine.

procedure DoIt(Const A : Integer; var B : Integer);

begin

B := A * 2;

ShowMessage(Format('A in the procedure = %d',[A]));

end;

procedure TForm1.InitializeForm;

var

A,B : Integer;

begin

inherited;

// this is a good place to initialize components

A := 22;

// B is not defined here

// Call the procedure

DoIt(A,B);

ShowMessage(Format('B has been set to = %d',[B]));

end;

Notice that when defining two argument types, the arguments are separated with a ;.

function InitializeForm(Self) {

var A$9 = 0;

TW3CustomForm.InitializeForm(Self);

A$9 = 22;

alert(("A in program before call = "+A$9.toString()));

DoIt(A$9);

alert(("A in program now = "+A$10.toString()));

}

function InitializeForm(Self) {

var A$9 = {};

A$9.v = 0;

TW3CustomForm.InitializeForm(Self);

A$9.v = 22;

alert(("A in program before call = "+A$9.v.toString()));

DoIt(A$9);

alert(("A in program now = "+A$9.v.toString()));

}

function InitializeForm(Self) {

var A$9 = 0;

var B$3 = {};

B$3.v = 0;

TW3CustomForm.InitializeForm(Self);

A$9 = 22;

DoIt(A$9,B$3);

alert(("B has been set to = "+B$3.v.toString()));

}

One of the benefits of Object Oriented programming is that some of the rigidity of procedural languages was relaxed. This has spilled over into non object orientation subroutines (as opposed to class methods).
One of the benefits is that we can define two or more subroutines that have exactly the same name. Pas2JS is able to tell them apart by the different number or types of parameters.
The example below illustrates this with two versions of the DoIt procedure using Pas2JS.

procedure DoIt; overload;

begin

ShowMessage('DoIt with no parameters called');

end;

procedure DoIt(msg : String); overload;

begin

ShowMessage('DoIt called with parameter : '+msg);

end;

procedure TForm1.InitializeForm;

begin

inherited;

// Call the procedure using no parameters

DoIt;

// Now call the procedure using one parameter

DoIt('Hi there');

end;