Free Pascal :

Reference guide.
Reference guide for Free Pascal, version 2.0.0
Document version 2.0
May 2005

Michaël Van Canneyt

Contents

1 Pascal Tokens 9
1.1 Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2 Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3 Reserved words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3.1 Turbo Pascal reserved words . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3.2 Delphi reserved words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3.3 Free Pascal reserved words . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3.4 Modifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.4 Identifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.5 Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.6 Labels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.7 Character strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2 Constants 14
2.1 Ordinary constants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2 Typed constants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3 Resource strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

3 Types 17
3.1 Base types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.1.1 Ordinal types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Integers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Boolean types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Enumeration types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Subrange types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1.2 Real types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.2 Character types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2.1 Char . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2.2 Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2.3 Short strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2.4 Ansistrings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

1
 CONTENTS

3.2.5 WideStrings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2.6 Constant strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2.7 PChar - Null terminated strings . . . . . . . . . . . . . . . . . . . . . . . . 25
3.3 Structured Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.3.1 Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Static arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Dynamic arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.3.2 Record types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.3.3 Set types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.3.4 File types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.4 Pointers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.5 Forward type declarations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.6 Procedural types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.7 Variant types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.7.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.7.2 Variants in assignments and expressions . . . . . . . . . . . . . . . . . . . . 39
3.7.3 Variants and interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

4 Variables 41
4.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.2 Declaration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.3 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.4 Thread Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.5 Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

5 Objects 47
5.1 Declaration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.2 Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.3 Constructors and destructors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.4 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.5 Method invocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Static methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Virtual methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Abstract methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.6 Visibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

6 Classes 54
6.1 Class definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
6.2 Class instantiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
6.3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
6.3.1 invocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

2
 CONTENTS

6.3.2 Virtual methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

6.3.3 Class methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.3.4 Message methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.4 Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

7 Interfaces 63
7.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
7.2 Interface identification: A GUID . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
7.3 Interfaces and COM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
7.4 CORBA and other Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

8 Expressions 67
8.1 Expression syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
8.2 Function calls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
8.3 Set constructors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
8.4 Value typecasts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
8.5 The @ operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
8.6 Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
8.6.1 Arithmetic operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
8.6.2 Logical operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
8.6.3 Boolean operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
8.6.4 String operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
8.6.5 Set operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
8.6.6 Relational operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

9 Statements 76
9.1 Simple statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
9.1.1 Assignments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
9.1.2 Procedure statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
9.1.3 Goto statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
9.2 Structured statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
9.2.1 Compound statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
9.2.2 The Case statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
9.2.3 The If..then..else statement . . . . . . . . . . . . . . . . . . . . . . 80
9.2.4 The For..to/downto..do statement . . . . . . . . . . . . . . . . . . . 81
9.2.5 The Repeat..until statement . . . . . . . . . . . . . . . . . . . . . . . 82
9.2.6 The While..do statement . . . . . . . . . . . . . . . . . . . . . . . . . . 83
9.2.7 The With statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
9.2.8 Exception Statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
9.3 Assembler statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

3
 CONTENTS

10 Using functions and procedures 87

10.1 Procedure declaration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
10.2 Function declaration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
10.3 Parameter lists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
10.3.1 Value parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
10.3.2 Variable parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
10.3.3 Out parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
10.3.4 Constant parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
10.3.5 Open array parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
10.3.6 Array of const . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
10.4 Function overloading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
10.5 Forward defined functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
10.6 External functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
10.7 Assembler functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
10.8 Modifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
10.8.1 alias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
10.8.2 cdecl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
10.8.3 export . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
10.8.4 inline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
10.8.5 interrupt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
10.8.6 pascal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
10.8.7 popstack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
10.8.8 public . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
10.8.9 register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
10.8.10 saveregisters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
10.8.11 safecall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
10.8.12 softfloat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
10.8.13 stdcall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
10.8.14 varargs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
10.9 Unsupported Turbo Pascal modifiers . . . . . . . . . . . . . . . . . . . . . . . . . . 100

11 Operator overloading 101

11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
11.2 Operator declarations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
11.3 Assignment operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
11.4 Arithmetic operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
11.5 Comparision operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

12 Programs, units, blocks 107

12.1 Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
12.2 Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

4
 CONTENTS

12.3 Blocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

12.4 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
12.4.1 Block scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
12.4.2 Record scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
12.4.3 Class scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
12.4.4 Unit scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
12.5 Libraries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

13 Exceptions 113
13.1 The raise statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
13.2 The try...except statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
13.3 The try...finally statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
13.4 Exception handling nesting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
13.5 Exception classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

14 Using assembler 117

14.1 Assembler statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
14.2 Assembler procedures and functions . . . . . . . . . . . . . . . . . . . . . . . . . . 117

5
List of Tables

3.1 Predefined integer types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

3.2 Predefined integer types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.3 Boolean types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.4 Supported Real types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.5 PChar pointer arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.6 Set Manipulation operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

8.1 Precedence of operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

8.2 Binary arithmetic operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
8.3 Unary arithmetic operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
8.4 Logical operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
8.5 Boolean operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
8.6 Set operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
8.7 Relational operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

9.1 Allowed C constructs in Free Pascal . . . . . . . . . . . . . . . . . . . . . . . . . . 77

10.1 Unsupported modifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

6
 LIST OF TABLES

About this guide

This document serves as the reference for the Pascal langauge as implemented by the Free Pascal
compiler. It describes all Pascal constructs supported by Free Pascal, and lists all supported data
types. It does not, however, give a detailed explanation of the pascal language. The aim is to list
which Pascal constructs are supported, and to show where the Free Pascal implementation differs
from the Turbo Pascal or Delphi implementations.
Earlier versions of this document also contained the reference documentation of the system unit and
objpas unit. This has been moved to the RTL reference guide.

Notations
Throughout this document, we will refer to functions, types and variables with typewriter font.
Functions and procedures have their own subsections, and for each function or procedure we have
the following topics:

Declaration The exact declaration of the function.

Description What does the procedure exactly do ?
Errors What errors can occur.
See Also Cross references to other related functions/commands.

The cross-references come in two flavours:

• References to other functions in this manual. In the printed copy, a number will appear after
this reference. It refers to the page where this function is explained. In the on-line help pages,
this is a hyperlink, which can be clicked to jump to the declaration.
• References to Unix manual pages. (For linux and unix related things only) they are printed in
typewriter font, and the number after it is the Unix manual section.

Syntax diagrams
All elements of the pascal language are explained in syntax diagrams. Syntax diagrams are like flow
charts. Reading a syntax diagram means getting from the left side to the right side, following the
arrows. When the right side of a syntax diagram is reached, and it ends with a single arrow, this
means the syntax diagram is continued on the next line. If the line ends on 2 arrows pointing to each
other, then the diagram is ended.
Syntactical elements are written like this
-
- syntactical elements are like this -

Keywords which must be typed exactly as in the diagram:

-
- keywords are like this -

When something can be repeated, there is an arrow around it:

-
- this can be repeated -
6

When there are different possibilities, they are listed in columns:

-
- First possibility -
Second possibility

7
 LIST OF TABLES

Note, that one of the possibilities can be empty:

-
- -
First possibility
Second possibility

This means that both the first or second possibility are optional. Of course, all these elements can be
combined and nested.

8
Chapter 1

Pascal Tokens

In this chapter we describe all the pascal reserved words, as well as the various ways to denote strings,
numbers, identifiers etc.

1.1 Symbols
Free Pascal allows all characters, digits and some special ASCII symbols in a Pascal source file.

Recognised symbols

-
- letter A...Z -
a...z

-
- digit 0...9 -

-
- hex digit 0...9 -
A...F
a...f

The following characters have a special meaning:

+ - * / = < > [ ] . , ( ) : ^ @ { } $ #

and the following character pairs too:

<= >= := += -= *= /= (* *) (. .) //

When used in a range specifier, the character pair (. is equivalent to the left square bracket [.
Likewise, the character pair .) is equivalent to the right square bracket ]. When used for comment
delimiters, the character pair (* is equivalent to the left brace { and the character pair *) is equiva-
lent to the right brace }. These character pairs retain their normal meaning in string expressions.

1.2 Comments
Free Pascal supports the use of nested comments. The following constructs are valid comments:

9
 CHAPTER 1. PASCAL TOKENS

(* This is an old style comment *)

{ This is a Turbo Pascal comment }
// This is a Delphi comment. All is ignored till the end of the line.

The following are valid ways of nesting comments:

{ Comment 1 (* comment 2 *) }
(* Comment 1 { comment 2 } *)
{ comment 1 // Comment 2 }
(* comment 1 // Comment 2 *)
// comment 1 (* comment 2 *)
// comment 1 { comment 2 }

The last two comments must be on one line. The following two will give errors:

// Valid comment { No longer valid comment !!

}

and

// Valid comment (* No longer valid comment !!

*)

The compiler will react with a ’invalid character’ error when it encounters such constructs, regardless
of the -So switch.

1.3 Reserved words

Reserved words are part of the Pascal language, and cannot be redefined. They will be denoted as
this throughout the syntax diagrams. Reserved words can be typed regardless of case, i.e. Pascal is
case insensitive. We make a distinction between Turbo Pascal and Delphi reserved words, since with
the -So switch, only the Turbo Pascal reserved words are recognised, and the Delphi ones can be
redefined. By default, Free Pascal recognises the Delphi reserved words.

1.3.1 Turbo Pascal reserved words

The following keywords exist in Turbo Pascal mode

absolute else nil shl

and end not shr
array file object string
asm for of then
begin function on to
break goto operator type
case if or unit
const implementation packed until
constructor in procedure uses
continue inherited program var
destructor inline record while
div interface repeat with
do label self xor
downto mod set

10
 CHAPTER 1. PASCAL TOKENS

1.3.2 Delphi reserved words

The Delphi (II) reserved words are the same as the pascal ones, plus the following ones:

as finalization library threadvar

class finally on try
except initialization property
exports is raise

1.3.3 Free Pascal reserved words

On top of the Turbo Pascal and Delphi reserved words, Free Pascal also considers the following as
reserved words:

dispose false true

exit new

1.3.4 Modifiers
The following is a list of all modifiers. They are not exactly reserved words in the sense that they can
be used as identifiers, but in specific places, they have a special meaning for the compiler.

absolute far pascal safecall

abstract far16 popstack saveregisters
alias forward private softfloat
assembler fpccall protected stdcall
cdecl index public virtual
default name published write
export near read
external override register

Remark: Predefined types such as Byte, Boolean and constants such as maxint are not reserved words.
They are identifiers, declared in the system unit. This means that these types can be redefined in other
units. The programmer is, however, not encouraged to do this, as it will cause a lot of confusion.

1.4 Identifiers
Identifiers denote constants, types, variables, procedures and functions, units, and programs. All
names of things that are defined are identifiers. An identifier consists of 255 significant characters
(letters, digits and the underscore character), from which the first must be an alphanumeric character,
or an underscore (_) The following diagram gives the basic syntax for identifiers.

Identifiers

-
- identifier letter -
_ 6 letter
digit
_

11
 CHAPTER 1. PASCAL TOKENS

1.5 Numbers
Numbers are by default denoted in decimal notation. Real (or decimal) numbers are written using
engineering or scientific notation (e.g. 0.314E1).
For integer type constants, Free Pascal supports 4 formats:

1. Normal, decimal format (base 10). This is the standard format.

2. Hexadecimal format (base 16), in the same way as Turbo Pascal does. To specify a constant
value in hexadecimal format, prepend it with a dollar sign ($). Thus, the hexadecimal $FF
equals 255 decimal. Note that case is insignificant when using hexadecimal constants.
3. As of version 1.0.7, Octal format (base 8) is also supported. To specify a constant in octal
format, prepend it with a ampersand (&). For instance 15 is specified in octal notation as &17.
4. Binary notation (base 2). A binary number can be specified by preceding it with a percent sign
(%). Thus, 255 can be specified in binary notation as %11111111.

The following diagrams show the syntax for numbers.

Numbers

-
- hex digit sequence hex digit -
6

-
- octal digit sequence octal digit -
6

-
- bin digit sequence 1 -
60

-
- digit sequence digit -
6

-
- unsigned integer digit sequence -
$ hex digit sequence
% bin digit sequence
- sign +
- -
-
-
- unsigned real digit sequence -
. digit sequence scale factor
-
- scale factor E digit sequence -
e sign
-
- unsigned number unsigned real -
unsigned integer
-
- signed number unsigned number -
sign

Remark: It is to note that all decimal constants which do no fit within the -2147483648..2147483647 range,
are silently and automatically parsed as 64-bit integer constants as of version 1.9.0. Earliers versions
would convert it to a real-typed constant.
Remark: Note that Octal and Binary notation are not supported in TP or Delphi compatibility mode.

12
 CHAPTER 1. PASCAL TOKENS

1.6 Labels
Labels can be digit sequences or identifiers.

Label

-
- label digit sequence -
identifier

Remark: Note that the -Sg switch must be specified before labels can be used. By default, Free Pascal doesn’t
support label and goto statements.

1.7 Character strings

A character string (or string for short) is a sequence of zero or more characters from the ASCII
character set, enclosed by single quotes, and on 1 line of the program source. A character set with
nothing between the quotes (’’) is an empty string.

Character strings

-
- character string quoted string -
6 control string

-
- quoted string ’ string character ’ -
6

-
- string character Any character except ’ or CR -
”

-
- control string # unsigned integer -
6

13
Chapter 2

Constants

Just as in Turbo Pascal, Free Pascal supports both normal and typed constants.

2.1 Ordinary constants

Ordinary constants declarations are not different from the Turbo Pascal or Delphi implementation.

Constant declaration

-
- constant declaration identifier = expression ; -
6

The compiler must be able to evaluate the expression in a constant declaration at compile time. This
means that most of the functions in the Run-Time library cannot be used in a constant declaration.
Operators such as +, -, *, /, not, and, or, div, mod, ord, chr, sizeof, pi,
int, trunc, round, frac, odd can be used, however. For more information on expres-
sions, see chapter 8, page 67. Only constants of the following types can be declared: Ordinal
types, Real types, Char, and String. The following are all valid constant declarations:

Const
e = 2.7182818; { Real type constant. }
a = 2; { Ordinal (Integer) type constant. }
c = ’4’; { Character type constant. }
s = ’This is a constant string’; {String type constant.}
s = chr(32)
ls = SizeOf(Longint);

Assigning a value to an ordinary constant is not permitted. Thus, given the previous declaration, the
following will result in a compiler error:

s := ’some other string’;

Prior to version 1.9, Free Pascal did not correctly support 64-bit constants. As of version 1.9, 64-bits
constants can be specified.

14
 CHAPTER 2. CONSTANTS

2.2 Typed constants

Typed constants serve to provide a program with initialised variables. Contrary to ordinary constants,
they may be assigned to at run-time. The difference with normal variables is that their value is
initialised when the program starts, whereas normal variables must be initialised explicitly.

Typed constant declaration

-
- typed constant declaration identifier : type = typed constant ; -
6

-
- typed constant constant -
address constant
array constant
record constant
procedural constant

Given the declaration:

Const
S : String = ’This is a typed constant string’;

The following is a valid assignment:

S := ’Result : ’+Func;

Where Func is a function that returns a String. Typed constants are often used to initialize arrays
and records. For arrays, the initial elements must be specified, surrounded by round brackets, and
separated by commas. The number of elements must be exactly the same as the number of elements
in the declaration of the type. As an example:

Const
tt : array [1..3] of string[20] = (’ikke’, ’gij’, ’hij’);
ti : array [1..3] of Longint = (1,2,3);

For constant records, each element of the record should be specified, in the form Field : Value,
separated by commas, and surrounded by round brackets. As an example:

Type
Point = record
X,Y : Real
end;
Const
Origin : Point = (X:0.0; Y:0.0);

The order of the fields in a constant record needs to be the same as in the type declaration, otherwise
a compile-time error will occur.
Remark: It should be stressed that typed constants are initialized at program start. This is also true for
local typed constants. Local typed constants are also initialized at program start. If their value was
changed during previous invocations of the function, they will retain their changed value, i.e. they
are not initialized each time the function is invoked.

15
 CHAPTER 2. CONSTANTS

2.3 Resource strings

A special kind of constant declaration part is the Resourestring part. This part is like a Const
section, but it only allows to declare constant of type string. This part is only available in the Delphi
or objfpc mode.
The following is an example of a resourcestring definition:

Resourcestring

FileMenu = ’&File...’;
EditMenu = ’&Edit...’;

All string constants defined in the resourcestring section are stored in special tables, allowing to
manipulate the values of the strings at runtime with some special mechanisms.
Semantically, the strings are like constants; Values can not be assigned to them, except through the
special mechanisms in the objpas unit. However, they can be used in assignments or expressions
as normal constants. The main use of the resourcestring section is to provide an easy means of
internationalization.
More on the subject of resourcestrings can be found in the Programmers guide, and in the chapter on
the objpas later in this manual.

16
Chapter 3

Types

All variables have a type. Free Pascal supports the same basic types as Turbo Pascal, with some
extra types from Delphi. The programmer can declare his own types, which is in essence defining an
identifier that can be used to denote this custom type when declaring variables further in the source
code.

Type declaration

-
- type declaration identifier = type ; -

There are 7 major type classes :

Types

-
- type simple type -
string type
structured type
pointer type
procedural type
type identifier

The last class, type identifier, is just a means to give another name to a type. This presents a way
to make types platform independent, by only using these types, and then defining these types for
each platform individually. The programmer that uses these units doesn’t have to worry about type
size and so on. It also allows to use shortcut names for fully qualified type names. e.g. define
[Link] as Olongint and then redefine longint.

3.1 Base types

The base or simple types of Free Pascal are the Delphi types. We will discuss each separate.

Simple types

17
 CHAPTER 3. TYPES

-
- simple type ordinal type -
real type
-
- real type real type identifier -

3.1.1 Ordinal types

With the exception of int64, qword and Real types, all base types are ordinal types. Ordinal types
have the following characteristics:

1. Ordinal types are countable and ordered, i.e. it is, in principle, possible to start counting them
one bye one, in a specified order. This property allows the operation of functions as Inc (??),
Ord (??), Dec (??) on ordinal types to be defined.
2. Ordinal values have a smallest possible value. Trying to apply the Pred (??) function on the
smallest possible value will generate a range check error if range checking is enabled.
3. Ordinal values have a largest possible value. Trying to apply the Succ (??) function on the
largest possible value will generate a range check error if range checking is enabled.

Integers

A list of pre-defined integer types is presented in table (3.1) The integer types, and their ranges and

Table 3.1: Predefined integer types

Name
Integer
Shortint
SmallInt
Longint
Longword
Int64
Byte
Word
Cardinal
QWord
Boolean
ByteBool
LongBool
Char

sizes, that are predefined in Free Pascal are listed in table (3.2). It is to note that the qword and
int64 types are not true ordinals, so some pascal constructs will not work with these two integer
types.
The integer type maps to the smallint type in the default Free Pascal mode. It maps to either a
longint or int64 in either Delphi or ObjFPC mode. The cardinal type is currently always mapped
to the longword type. The definition of the cardinal and integer types may change from one
architecture to another and from one compiler mode to another. They usually have the same size as
the underlying target architecture.

18
 CHAPTER 3. TYPES

Table 3.2: Predefined integer types

Type Range Size in bytes

Byte 0 .. 255 1
Shortint -128 .. 127 1
Smallint -32768 .. 32767 2
Word 0 .. 65535 2
Integer either smallint, longint or int64 size 2,4 or 8
Cardinal either word, longword or qword size 2,4 or 8
Longint -2147483648 .. 2147483647 4
Longword 0..4294967295 4
Int64 -9223372036854775808 .. 9223372036854775807 8
QWord 0 .. 18446744073709551615 8

Free Pascal does automatic type conversion in expressions where different kinds of integer types are
used.

Boolean types

Free Pascal supports the Boolean type, with its two pre-defined possible values True and False.
It also supports the ByteBool, WordBool and LongBool types. These are the only two values
that can be assigned to a Boolean type. Of course, any expression that resolves to a boolean
value, can also be assigned to a boolean type. Assuming B to be of type Boolean, the following

Table 3.3: Boolean types

Name Size Ord(True)

Boolean 1 1
ByteBool 1 Any nonzero value
WordBool 2 Any nonzero value
LongBool 4 Any nonzero value

are valid assignments:

B := True;
B := False;
B := 1<>2; { Results in B := True }

Boolean expressions are also used in conditions.

Remark: In Free Pascal, boolean expressions are always evaluated in such a way that when the result is known,
the rest of the expression will no longer be evaluated (Called short-cut evaluation). In the following
example, the function Func will never be called, which may have strange side-effects.

...
B := False;
A := B and Func;

Here Func is a function which returns a Boolean type.

19
 CHAPTER 3. TYPES

Enumeration types

Enumeration types are supported in Free Pascal. On top of the Turbo Pascal implementation, Free
Pascal allows also a C-style extension of the enumeration type, where a value is assigned to a partic-
ular element of the enumeration list.

Enumerated types

-
- enumerated type ( identifier list ) -
6 assigned enum list
,

-
- identifier list identifier -
6 ,

-
- assigned enum list identifier := expression -
6 ,

(see chapter 8, page 67 for how to use expressions) When using assigned enumerated types, the
assigned elements must be in ascending numerical order in the list, or the compiler will complain.
The expressions used in assigned enumerated elements must be known at compile time. So the
following is a correct enumerated type declaration:

Type
Direction = ( North, East, South, West );

The C style enumeration type looks as follows:

Type
EnumType = (one, two, three, forty := 40,fortyone);

As a result, the ordinal number of forty is 40, and not 3, as it would be when the ’:= 40’
wasn’t present. The ordinal value of fortyone is then 41, and not 4, as it would be when the
assignment wasn’t present. After an assignment in an enumerated definition the compiler adds 1 to
the assigned value to assign to the next enumerated value. When specifying such an enumeration
type, it is important to keep in mind that the enumerated elements should be kept in ascending order.
The following will produce a compiler error:

Type
EnumType = (one, two, three, forty := 40, thirty := 30);

It is necessary to keep forty and thirty in the correct order. When using enumeration types it is
important to keep the following points in mind:

1. The Pred and Succ functions cannot be used on this kind of enumeration types. Trying to
do this anyhow will result in a compiler error.
2. Enumeration types stored using a default size. This behaviour can be changed with the {$PACKENUM
n} compiler directive, which tells the compiler the minimal number of bytes to be used for
enumeration types. For instance

20
 CHAPTER 3. TYPES

Type
{$PACKENUM 4}
LargeEnum = ( BigOne, BigTwo, BigThree );
{$PACKENUM 1}
SmallEnum = ( one, two, three );
Var S : SmallEnum;
L : LargeEnum;
begin
WriteLn (’Small enum : ’,SizeOf(S));
WriteLn (’Large enum : ’,SizeOf(L));
end.
will, when run, print the following:
Small enum : 1
Large enum : 4

More information can be found in the Programmers guide, in the compiler directives section.

Subrange types

A subrange type is a range of values from an ordinal type (the host type). To define a subrange type,
one must specify it’s limiting values: the highest and lowest value of the type.

Subrange types

-
- subrange type constant .. constant -

Some of the predefined integer types are defined as subrange types:

Type
Longint = $80000000..$7fffffff;
Integer = -32768..32767;
shortint = -128..127;
byte = 0..255;
Word = 0..65535;

Subrange types of enumeration types can also be defined:

Type
Days = (monday,tuesday,wednesday,thursday,friday,
saturday,sunday);
WorkDays = monday .. friday;
WeekEnd = Saturday .. Sunday;

3.1.2 Real types

Free Pascal uses the math coprocessor (or emulation) for all its floating-point calculations. The Real
native type is processor dependant, but it is either Single or Double. Only the IEEE floating point
types are supported, and these depend on the target processor and emulation options. The true Turbo
Pascal compatible types are listed in table (3.4). The Comp type is, in effect, a 64-bit integer and
is not available on all target platforms. To get more information on the supported types for each
platform, refer to the Programmers guide.

21
 CHAPTER 3. TYPES

Table 3.4: Supported Real types

Type Range Significant digits Size

Real platform dependant ??? 4 or 8
Single 1.5E-45 .. 3.4E38 7-8 4
Double 5.0E-324 .. 1.7E308 15-16 8
Extended 1.9E-4951 .. 1.1E4932 19-20 10
Comp -2E64+1 .. 2E63-1 19-20 8

3.2 Character types

3.2.1 Char
Free Pascal supports the type Char. A Char is exactly 1 byte in size, and contains one character.
A character constant can be specified by enclosing the character in single quotes, as follows : ’a’ or
’A’ are both character constants. A character can also be specified by its ASCII value, by preceding
the ASCII value with the number symbol (#). For example specifying #65 would be the same as
’A’. Also, the caret character (^) can be used in combination with a letter to specify a character
with ASCII value less than 27. Thus ^G equals #7 (G is the seventh letter in the alphabet.) When
the single quote character must be represented, it should be typed two times successively, thus ””
represents the single quote character.

3.2.2 Strings
Free Pascal supports the String type as it is defined in Turbo Pascal (A sequence of characters with
a specified length) and it supports ansistrings as in Delphi. To declare a variable as a string, use the
following type specification:

ShortString

-
- string type string -
[ unsigned integer ]

The meaning of a string declaration statement is interpreted differently depending on the {$H}
switch. The above declaration can declare an ansistrng or a short string.
Whatever the actual type, ansistrings and short strings can be used interchangeably. The compiler
always takes care of the necessary type conversions. Note, however, that the result of an expression
that contains ansistrings and short strings will always be an ansistring.

3.2.3 Short strings

A string declaration declares a short string in the following cases:

1. If the switch is off: {$H-}, the string declaration will always be a short string declaration.
2. If the switch is on {$H+}, and there is a length specifier, the declaration is a short string
declaration.

The predefined type ShortString is defined as a string of length 255:

22
 CHAPTER 3. TYPES

ShortString = String[255];

If the size of the string is not specified, 255 is taken as a default. The length of the string can be
obtained with the Length (??) standard runtime routine. For example in

{$H-}

Type
NameString = String[10];
StreetString = String;

NameString can contain a maximum of 10 characters. While StreetString can contain up to

255 characters.

3.2.4 Ansistrings
Ansistrings are strings that have no length limit. They are reference counted and null terminated.
Internally, an ansistring is treated as a pointer. This is all handled transparantly, i.e. they can be
manipulated as a normal short string. Ansistrings can be defined using the predefined AnsiString
type.
If the {$H} switch is on, then a string definition using the regular String keyword and that doesn’t
contain a length specifier, will be regarded as an ansistring as well. If a length specifier is present, a
short string will be used, regardless of the {$H} setting.
If the string is empty (”), then the internal pointer representation of the string pointer is Nil. If the
string is not empty, then the pointer points to a structure in heap memory.
The internal representation as a pointer, and the automatic null-termination make it possible to type-
cast an ansistring to a pchar. If the string is empty (so the pointer is nil) then the compiler makes sure
that the typecasted pchar will point to a null byte.
Assigning one ansistring to another doesn’t involve moving the actual string. A statement

S2:=S1;

results in the reference count of S2 being decreased by one, The referece count of S1 is increased by
one, and finally S1 (as a pointer) is copied to S2. This is a significant speed-up in the code.
If the reference count reaches zero, then the memory occupied by the string is deallocated automati-
cally, so no memory leaks arise.
When an ansistring is declared, the Free Pascal compiler initially allocates just memory for a pointer,
not more. This pointer is guaranteed to be nil, meaning that the string is initially empty. This is true
for local and global ansistrings or anstrings that are part of a structure (arrays, records or objects).
This does introduce an overhead. For instance, declaring

Var
A : Array[1..100000] of string;

Will copy 100,000 times nil into A. When A goes out of scope, then the reference count of the
100,000 strings will be decreased by 1 for each of these strings. All this happens invisibly for the
programmer, but when considering performance issues, this is important.
Memory will be allocated only when the string is assigned a value. If the string goes out of scope,
then its reference count is automatically decreased by 1. If the reference count reaches zero, the
memory reserved for the string is released.
If a value is assigned to a character of a string that has a reference count greater than 1, such as in the
following statements:

23
 CHAPTER 3. TYPES

S:=T; { reference count for S and T is now 2 }

S[I]:=’@’;

then a copy of the string is created before the assignment. This is known as copy-on-write semantics.
The Length (??) function must be used to get the length of an ansistring.
To set the length of an ansistring, the SetLength (??) function must be used. Constant ansistrings
have a reference count of -1 and are treated specially.
Ansistrings are converted to short strings by the compiler if needed, this means that the use of an-
sistrings and short strings can be mixed without problems.
Ansistrings can be typecasted to PChar or Pointer types:

Var P : Pointer;
PC : PChar;
S : AnsiString;

begin
S :=’This is an ansistring’;
PC:=Pchar(S);
P :=Pointer(S);

There is a difference between the two typecasts. When an empty ansistring is typecasted to a pointer,
the pointer wil be Nil. If an empty ansistring is typecasted to a PChar, then the result will be a
pointer to a zero byte (an empty string).
The result of such a typecast must be used with care. In general, it is best to consider the result
of such a typecast as read-only, i.e. suitable for passing to a procedure that needs a constant pchar
argument.
It is therefore NOT advisable to typecast one of the following:

1. expressions.
2. strings that have reference count larger than 0. (call uniquestring to ensure a string has refer-
ence count 1)

3.2.5 WideStrings
Widestrings (used to represent unicode character strings) are implemented in much the same way
as ansistrings: reference counted, null-terminated arrays, only they are implemented as arrays of
WideChars instead of regular Chars. A WideChar is a two-byte character (an element of
a DBCS: Double Byte Character Set). Mostly the same rules apply for WideStrings as for
AnsiStrings. The compiler transparantly converts WideStrings to AnsiStrings and vice versa.
Similarly to the typecast of an Ansistring to a PChar null-terminated array of characters, a WideString
can be converted to a PWideChar null-terminated array of characters. Note that the PWideChar
array is terminated by 2 null bytes instead of 1, so a typecast to a pchar is not automatic.
The compiler itself provides no support for any conversion from Unicode to ansistrings or vice versa;
2 procedural variables are present in the system unit which can be set to handle the conversion. For
more information, see the system units reference.

3.2.6 Constant strings

To specify a constant string, it must be enclosed in single-quotes, just as a Char type, only now more
than one character is allowed. Given that S is of type String, the following are valid assignments:

24
 CHAPTER 3. TYPES

S := ’This is a string.’;
S := ’One’+’, Two’+’, Three’;
S := ’This isn’’t difficult !’;
S := ’This is a weird character : ’#145’ !’;

As can be seen, the single quote character is represented by 2 single-quote characters next to each
other. Strange characters can be specified by their ASCII value. The example shows also that two
strings can be added. The resulting string is just the concatenation of the first with the second string,
without spaces in between them. Strings can not be substracted, however.
Whether the constant string is stored as an ansistring or a short string depends on the settings of the
{$H} switch.

3.2.7 PChar - Null terminated strings

Free Pascal supports the Delphi implementation of the PChar type. PChar is defined as a pointer to
a Char type, but allows additional operations. The PChar type can be understood best as the Pascal
equivalent of a C-style null-terminated string, i.e. a variable of type PChar is a pointer that points
to an array of type Char, which is ended by a null-character (#0). Free Pascal supports initializing
of PChar typed constants, or a direct assignment. For example, the following pieces of code are
equivalent:

program one;
var p : PChar;
begin
P := ’This is a null-terminated string.’;
WriteLn (P);
end.

Results in the same as

program two;
const P : PChar = ’This is a null-terminated string.’
begin
WriteLn (P);
end.

These examples also show that it is possible to write the contents of the string to a file of type
Text. The strings unit contains procedures and functions that manipulate the PChar type as in the
standard C library. Since it is equivalent to a pointer to a type Char variable, it is also possible to do
the following:

Program three;
Var S : String[30];
P : PChar;
begin
S := ’This is a null-terminated string.’#0;
P := @S[1];
WriteLn (P);
end.

This will have the same result as the previous two examples. Null-terminated strings cannot be added
as normal Pascal strings. If two PChar strings mustt be concatenated; the functions from the unit
strings must be used.

25
 CHAPTER 3. TYPES

However, it is possible to do some pointer arithmetic. The operators + and - can be used to do
operations on PChar pointers. In table (3.5), P and Q are of type PChar, and I is of type Longint.

Table 3.5: PChar pointer arithmetic

Operation Result
P + I Adds I to the address pointed to by P.
I + P Adds I to the address pointed to by P.
P - I Substracts I from the address pointed to by P.
P - Q Returns, as an integer, the distance between 2 addresses
(or the number of characters between P and Q)

3.3 Structured Types

A structured type is a type that can hold multiple values in one variable. Stuctured types can be
nested to unlimited levels.

Structured Types

-
- structured type array type -
record type
object type
class type
class reference type
interface type
set type
file type

Unlike Delphi, Free Pascal does not support the keyword Packed for all structured types, as can be
seen in the syntax diagram. It will be mentioned when a type supports the packed keyword. In the
following, each of the possible structured types is discussed.

3.3.1 Arrays
Free Pascal supports arrays as in Turbo Pascal, multi-dimensional arrays and packed arrays are also
supported, as well as the dynamic arrays of Delphi:

Array types

-
- array type array of type -
packed [ ordinal type ]
6 ,

26
 CHAPTER 3. TYPES

Static arrays

When the range of the array is included in the array definition, it is called a static array. Trying to
access an element with an index that is outside the declared range will generate a run-time error (if
range checking is on). The following is an example of a valid array declaration:

Type
RealArray = Array [1..100] of Real;

Valid indexes for accessing an element of the array are between 1 and 100, where the borders 1 and
100 are included. As in Turbo Pascal, if the array component type is in itself an array, it is possible
to combine the two arrays into one multi-dimensional array. The following declaration:

Type
APoints = array[1..100] of Array[1..3] of Real;

is equivalent to the following declaration:

Type
APoints = array[1..100,1..3] of Real;

The functions High (??) and Low (??) return the high and low bounds of the leftmost index type of
the array. In the above case, this would be 100 and 1.
When static array-type variables are assigned to each other, the contents of the whole array is copied.
This is also true for multi-dimensional arrays:

program testarray1;

Type
TA = Array[0..9,0..9] of Integer;

var
A,B : TA;
I,J : Integer;
begin
For I:=0 to 9 do
For J:=0 to 9 do
A[I,J]:=I*J;
For I:=0 to 9 do
begin
For J:=0 to 9 do
Write(A[I,J]:2,’ ’);
Writeln;
end;
B:=A;
Writeln;
For I:=0 to 9 do
For J:=0 to 9 do
A[9-I,9-J]:=I*J;
For I:=0 to 9 do
begin
For J:=0 to 9 do
Write(B[I,J]:2,’ ’);
Writeln;

27
 CHAPTER 3. TYPES

end;
end.

The output will be 2 identical matrices.

Dynamic arrays

As of version 1.1, Free Pascal also knows dynamic arrays: In that case, the array range is omitted, as
in the following example:

Type
TByteArray : Array of Byte;

When declaring a variable of a dynamic array type, the initial length of the array is zero. The actual
length of the array must be set with the standard SetLength function, which will allocate the
memory to contain the array elements on the heap. The following example will set the length to
1000:

Var
A : TByteArray;

begin
SetLength(A,1000);

After a call to SetLength, valid array indexes are 0 to 999: the array index is always zero-based.
Note that the length of the array is set in elements, not in bytes of allocated memory (although these
may be the same). The amount of memory allocated is the size of the array multiplied by the size
of 1 element in the array. The memory will be disposed of at the exit of the current procedure or
function.
It is also possible to resize the array: in that case, as much of the elements in the array as will fit in
the new size, will be kept. The array can be resized to zero, which effectively resets the variable.
At all times, trying to access an element of the array that is not in the current length of the array will
generate a run-time error.
Assignment of one dynamic array-type variable to another will let both variables point to the same
array. Contrary to ansistrings, an assignment to an element of one array will be reflected in the other:

Var
A,B : TByteArray;

begin
SetLength(A,10);
A[1]:=33;
B:=A;
A[1]:=31;

After the second assignment, the first element in B will also contain 31.
It can also be seen from the output of the following example:

program testarray1;

Type
TA = Array of array of Integer;

28
 CHAPTER 3. TYPES

var
A,B : TA;
I,J : Integer;
begin
Setlength(A,10,10);
For I:=0 to 9 do
For J:=0 to 9 do
A[I,J]:=I*J;
For I:=0 to 9 do
begin
For J:=0 to 9 do
Write(A[I,J]:2,’ ’);
Writeln;
end;
B:=A;
Writeln;
For I:=0 to 9 do
For J:=0 to 9 do
A[9-I,9-J]:=I*J;
For I:=0 to 9 do
begin
For J:=0 to 9 do
Write(B[I,J]:2,’ ’);
Writeln;
end;
end.

The output will be a matrix of numbers, and then the same matrix, mirrorred.
Dynamic arrays are reference counted: if in one of the previous examples A goes out of scope and B
does not, then the array is not yet disposed of: the reference count of A (and B) is decreased with 1.
As soon as the reference count reaches zero, the memory is disposed of.
It is also possible to copy and/or resize the array with the standard Copy function, which acts as the
copy function for strings:

program testarray3;

Type
TA = array of Integer;

var
A,B : TA;
I : Integer;

begin
Setlength(A,10);
For I:=0 to 9 do
A[I]:=I;
B:=Copy(A,3,9);
For I:=0 to 5 do
Writeln(B[I]);
end.

29
 CHAPTER 3. TYPES

The Copy function will copy 9 elements of the array to a new array. Starting at the element at index
3 (i.e. the fourth element) of the array.
The Low function on a dynamic array will always return 0, and the High function will return the
value Length-1, i.e., the value of the highest allowed array index. The Length function will
return the number of elements in the array.

3.3.2 Record types

Free Pascal supports fixed records and records with variant parts. The syntax diagram for a record
type is

Record types

-
- record type record end -
packed field list

-
- field list fixed fields -
variant part ;
fixed fields ;

-
- fixed fields identifier list : type -
6 ;

-
- variant part case ordinal type identifier of variant -
identifier : 6 ;

-
- variant constant , : ( ) -
6 field list

So the following are valid record types declarations:

Type
Point = Record
X,Y,Z : Real;
end;
RPoint = Record
Case Boolean of
False : (X,Y,Z : Real);
True : (R,theta,phi : Real);
end;
BetterRPoint = Record
Case UsePolar : Boolean of
False : (X,Y,Z : Real);
True : (R,theta,phi : Real);
end;

The variant part must be last in the record. The optional identifier in the case statement serves to
access the tag field value, which otherwise would be invisible to the programmer. It can be used to
see which variant is active at a certain time. In effect, it introduces a new field in the record.
Remark: It is possible to nest variant parts, as in:

30
 CHAPTER 3. TYPES

Type
MyRec = Record
X : Longint;
Case byte of
2 : (Y : Longint;
case byte of
3 : (Z : Longint);
);
end;

The size of a record is the sum of the sizes of its fields, each size of a field is rounded up to a power of
two. If the record contains a variant part, the size of the variant part is the size of the biggest variant,
plus the size of the tag field type if an identifier was declared for it. Here also, the size of each
part is first rounded up to two. So in the above example, SizeOf (??) would return 24 for Point,
24 for RPoint and 26 for BetterRPoint. For MyRec, the value would be 12. If a typed file
with records, produced by a Turbo Pascal program, must be read, then chances are that attempting
to read that file correctly will fail. The reason for this is that by default, elements of a record are
aligned at 2-byte boundaries, for performance reasons. This default behaviour can be changed with
the {$PackRecords n} switch. Possible values for n are 1, 2, 4, 16 or Default. This switch
tells the compiler to align elements of a record or object or class that have size larger than n on n byte
boundaries. Elements that have size smaller or equal than n are aligned on natural boundaries, i.e.
to the first power of two that is larger than or equal to the size of the record element. The keyword
Default selects the default value for the platform that the code is compiled for (currently, this is 2
on all platforms) Take a look at the following program:

Program PackRecordsDemo;
type
{$PackRecords 2}
Trec1 = Record
A : byte;
B : Word;
end;

{$PackRecords 1}
Trec2 = Record
A : Byte;
B : Word;
end;
{$PackRecords 2}
Trec3 = Record
A,B : byte;
end;

{$PackRecords 1}
Trec4 = Record
A,B : Byte;
end;
{$PackRecords 4}
Trec5 = Record
A : Byte;
B : Array[1..3] of byte;
C : byte;
end;

31
 CHAPTER 3. TYPES

{$PackRecords 8}
Trec6 = Record
A : Byte;
B : Array[1..3] of byte;
C : byte;
end;
{$PackRecords 4}
Trec7 = Record
A : Byte;
B : Array[1..7] of byte;
C : byte;
end;

{$PackRecords 8}
Trec8 = Record
A : Byte;
B : Array[1..7] of byte;
C : byte;
end;
Var rec1 : Trec1;
rec2 : Trec2;
rec3 : TRec3;
rec4 : TRec4;
rec5 : Trec5;
rec6 : TRec6;
rec7 : TRec7;
rec8 : TRec8;

begin
Write (’Size Trec1 : ’,SizeOf(Trec1));
Writeln (’ Offset B : ’,Longint(@rec1.B)-Longint(@rec1));
Write (’Size Trec2 : ’,SizeOf(Trec2));
Writeln (’ Offset B : ’,Longint(@rec2.B)-Longint(@rec2));
Write (’Size Trec3 : ’,SizeOf(Trec3));
Writeln (’ Offset B : ’,Longint(@rec3.B)-Longint(@rec3));
Write (’Size Trec4 : ’,SizeOf(Trec4));
Writeln (’ Offset B : ’,Longint(@rec4.B)-Longint(@rec4));
Write (’Size Trec5 : ’,SizeOf(Trec5));
Writeln (’ Offset B : ’,Longint(@rec5.B)-Longint(@rec5),
’ Offset C : ’,Longint(@rec5.C)-Longint(@rec5));
Write (’Size Trec6 : ’,SizeOf(Trec6));
Writeln (’ Offset B : ’,Longint(@rec6.B)-Longint(@rec6),
’ Offset C : ’,Longint(@rec6.C)-Longint(@rec6));
Write (’Size Trec7 : ’,SizeOf(Trec7));
Writeln (’ Offset B : ’,Longint(@rec7.B)-Longint(@rec7),
’ Offset C : ’,Longint(@rec7.C)-Longint(@rec7));
Write (’Size Trec8 : ’,SizeOf(Trec8));
Writeln (’ Offset B : ’,Longint(@rec8.B)-Longint(@rec8),
’ Offset C : ’,Longint(@rec8.C)-Longint(@rec8));
end.

The output of this program will be :

Size Trec1 : 4 Offset B : 2

32
 CHAPTER 3. TYPES

Size Trec2 : 3 Offset B : 1

Size Trec3 : 2 Offset B : 1
Size Trec4 : 2 Offset B : 1
Size Trec5 : 8 Offset B : 4 Offset C : 7
Size Trec6 : 8 Offset B : 4 Offset C : 7
Size Trec7 : 12 Offset B : 4 Offset C : 11
Size Trec8 : 16 Offset B : 8 Offset C : 15

And this is as expected. In Trec1, since B has size 2, it is aligned on a 2 byte boundary, thus leaving
an empty byte between A and B, and making the total size 4. In Trec2, B is aligned on a 1-byte
boundary, right after A, hence, the total size of the record is 3. For Trec3, the sizes of A,B are 1,
and hence they are aligned on 1 byte boundaries. The same is true for Trec4. For Trec5, since the
size of B – 3 – is smaller than 4, B will be on a 4-byte boundary, as this is the first power of two that
is larger than it’s size. The same holds for Trec6. For Trec7, B is aligned on a 4 byte boundary,
since it’s size – 7 – is larger than 4. However, in Trec8, it is aligned on a 8-byte boundary, since 8
is the first power of two that is greater than 7, thus making the total size of the record 16. Free Pascal
supports also the ’packed record’, this is a record where all the elements are byte-aligned. Thus the
two following declarations are equivalent:

{$PackRecords 1}
Trec2 = Record
A : Byte;
B : Word;
end;
{$PackRecords 2}

and

Trec2 = Packed Record

A : Byte;
B : Word;
end;

Note the {$PackRecords 2} after the first declaration !

3.3.3 Set types

Free Pascal supports the set types as in Turbo Pascal. The prototype of a set declaration is:

Set Types

-
- set type set of ordinal type -

Each of the elements of SetType must be of type TargetType. TargetType can be any ordinal
type with a range between 0 and 255. A set can contain maximally 255 elements. The following
are valid set declaration:

Type
Junk = Set of Char;

Days = (Mon, Tue, Wed, Thu, Fri, Sat, Sun);

WorkDays : Set of days;

33
 CHAPTER 3. TYPES

Given this set declarations, the following assignment is legal:

WorkDays := [ Mon, Tue, Wed, Thu, Fri];

The operators and functions for manipulations of sets are listed in table (3.6). Two sets can be

Table 3.6: Set Manipulation operators

Operation Operator
Union +
Difference -
Intersection *
Add element include
Delete element exclude

compared with the <> and = operators, but not (yet) with the < and > operators. The compiler stores
small sets (less than 32 elements) in a Longint, if the type range allows it. This allows for faster
processing and decreases program size. Otherwise, sets are stored in 32 bytes.

3.3.4 File types

File types are types that store a sequence of some base type, which can be any type except another file
type. It can contain (in principle) an infinite number of elements. File types are used commonly to
store data on disk. Nothing prevents the programmer, however, from writing a file driver that stores
it’s data in memory. Here is the type declaration for a file type:

File types

-
- file type file -
of type

If no type identifier is given, then the file is an untyped file; it can be considered as equivalent to a file
of bytes. Untyped files require special commands to act on them (see Blockread (??), Blockwrite
(??)). The following declaration declares a file of records:

Type
Point = Record
X,Y,Z : real;
end;
PointFile = File of Point;

Internally, files are represented by the FileRec record, which is declared in the DOS unit.
A special file type is the Text file type, represented by the TextRec record. A file of type Text
uses special input-output routines.

3.4 Pointers
Free Pascal supports the use of pointers. A variable of the pointer type contains an address in memory,
where the data of another variable may be stored.

34
 CHAPTER 3. TYPES

Pointer types

-
- pointer type ˆ type identifier -

As can be seen from this diagram, pointers are typed, which means that they point to a particular kind
of data. The type of this data must be known at compile time. Dereferencing the pointer (denoted by
adding ˆ after the variable name) behaves then like a variable. This variable has the type declared
in the pointer declaration, and the variable is stored in the address that is pointed to by the pointer
variable. Consider the following example:

Program pointers;
type
Buffer = String[255];
BufPtr = ^Buffer;
Var B : Buffer;
BP : BufPtr;
PP : Pointer;
etc..

In this example, BP is a pointer to a Buffer type; while B is a variable of type Buffer. B takes
256 bytes memory, and BP only takes 4 bytes of memory (enough to keep an adress in memory).
Remark: Free Pascal treats pointers much the same way as C does. This means that a pointer to some type
can be treated as being an array of this type. The pointer then points to the zeroeth element of this
array. Thus the following pointer declaration

Var p : ^Longint;

Can be considered equivalent to the following array declaration:

Var p : array[0..Infinity] of Longint;

The difference is that the former declaration allocates memory for the pointer only (not for the array),
and the second declaration allocates memory for the entire array. If the former is used, the memory
must be allocated manually, using the Getmem (??) function. The reference Pˆ is then the same as
p[0]. The following program illustrates this maybe more clear:

program PointerArray;
var i : Longint;
p : ^Longint;
pp : array[0..100] of Longint;
begin
for i := 0 to 100 do pp[i] := i; { Fill array }
p := @pp[0]; { Let p point to pp }
for i := 0 to 100 do
if p[i]<>pp[i] then
WriteLn (’Ohoh, problem !’)
end.

Free Pascal supports pointer arithmetic as C does. This means that, if P is a typed pointer, the
instructions

35
 CHAPTER 3. TYPES

Inc(P);
Dec(P);

Will increase, respectively decrease the address the pointer points to with the size of the type P is a
pointer to. For example

Var P : ^Longint;
...
Inc (p);

will increase P with 4. Normal arithmetic operators on pointers can also be used, that is, the following
are valid pointer arithmetic operations:

var p1,p2 : ^Longint;

L : Longint;
begin
P1 := @P2;
P2 := @L;
L := P1-P2;
P1 := P1-4;
P2 := P2+4;
end.

Here, the value that is added or substracted is multiplied by the size of the type the pointer points to.
In the previous example P1 will be decremented by 16 bytes, and P2 will be incremented by 16.

3.5 Forward type declarations

Programs often need to maintain a linked list of records. Each record then contains a pointer to the
next record (and possibly to the previous record as well). For type safety, it is best to define this
pointer as a typed pointer, so the next record can be allocated on the heap using the New call. In
order to do so, the record should be defined something like this:

Type
TListItem = Record
Data : Integer;
Next : ^TListItem;
end;

When trying to compile this, the compiler will complain that the TListItem type is not yet defined
when it encounters the Next declaration: This is correct, as the definition is still being parsed.
To be able to have the Next element as a typed pointer, a ’Forward type declaration’ must be intro-
duced:

Type
PListItem = ^TListItem;
TListItem = Record
Data : Integer;
Next : PTListItem;
end;

When the compiler encounters a typed pointer declaration where the referenced type is not yet known,
it postpones resolving the reference later on: The pointer definition is a ’Forward type declaration’.

36
 CHAPTER 3. TYPES

The referenced type should be introduced later in the same Type block. No other block may come
between the definition of the pointer type and the referenced type. Indeed, even the word Type
itself may not re-appear: in effect it would start a new type-block, causing the compiler to resolve
all pending declarations in the current block. In most cases, the definition of the referenced type will
follow immediatly after the definition of the pointer type, as shown in the above listing. The forward
defined type can be used in any type definition following its declaration.
Note that a forward type declaration is only possible with pointer types and classes, not with other
types.

3.6 Procedural types

Free Pascal has support for procedural types, although it differs a little from the Turbo Pascal imple-
mentation of them. The type declaration remains the same, as can be seen in the following syntax
diagram:

Procedural types

-
- procedural type function header -
procedure header of object ; call modifiers

-
- function header function formal parameter list : result type -

-
- procedure header procedure formal parameter list -

-
- call modifiers register -
cdecl
pascal
stdcall
safecall
saveregisters
popstack

For a description of formal parameter lists, see chapter 10, page 87. The two following examples are
valid type declarations:

Type TOneArg = Procedure (Var X : integer);

TNoArg = Function : Real;
var proc : TOneArg;
func : TNoArg;

One can assign the following values to a procedural type variable:

1. Nil, for both normal procedure pointers and method pointers.

2. A variable reference of a procedural type, i.e. another variable of the same type.
3. A global procedure or function address, with matching function or procedure header and call-
ing convention.
4. A method address.

Given these declarations, the following assignments are valid:

37
 CHAPTER 3. TYPES

Procedure printit (Var X : Integer);

begin
WriteLn (x);
end;
...
Proc := @printit;
Func := @Pi;

From this example, the difference with Turbo Pascal is clear: In Turbo Pascal it isn’t necessary to
use the address operator (@) when assigning a procedural type variable, whereas in Free Pascal it is
required (unless the -So switch is used, in which case the address operator can be dropped.)
Remark: The modifiers concerning the calling conventions must be the same as the declaration; i.e. the
following code would give an error:

Type TOneArgCcall = Procedure (Var X : integer);cdecl;

var proc : TOneArgCcall;
Procedure printit (Var X : Integer);
begin
WriteLn (x);
end;
begin
Proc := @printit;
end.

Because the TOneArgCcall type is a procedure that uses the cdecl calling convention.

3.7 Variant types

3.7.1 Definition
As of version 1.1, FPC has support for variants. For variant support to be enabled, the variants unit
must be included in every unit that uses variants in some way. Furthermore, the compiler must be in
Delphi or ObjFPC mode.
The type of a value stored in a variant is only determined at runtime: it depends what has been
assigned to the to the variant. Almost any type can be assigned to variants: ordinal types, string
types, int64 types. Structured types such as sets, records, arrays, files, objects and classes are not
assign-compatible with a variant, as well as pointers. Interfaces and COM or CORBA objects can be
assigned to a variant.
This means that the following assignments are valid:

Type
TMyEnum = (One,Two,Three);

Var
V : Variant;
I : Integer;
B : Byte;
W : Word;
Q : Int64;
E : Extended;
D : Double;
En : TMyEnum;

38
 CHAPTER 3. TYPES

AS : AnsiString;
WS : WideString;

begin
V:=I;
V:=B;
V:=W;
V:=Q;
V:=E;
V:=En;
V:=D:
V:=AS;
V:=WS;
end;

And of course vice-versa as well.

Remark: The enumerated type assignment is broken in the early 1.1 development series of the compiler. It is
expected that this is fixed soon.
A variant can hold an an array of values: All elements in the array have the same type (but can be of
type ’variant’). For a variant that contains an array, the variant can be indexed:

Program testv;

uses variants;

Var
A : Variant;
I : integer;

begin
A:=VarArrayCreate([1,10],varInteger);
For I:=1 to 10 do
A[I]:=I;
end.

(for the explanation of VarArrayCreate, see Unit reference.)

Note that when the array contains a string, this is not considered an ’array of characters’, and so the
variant cannot be indexed to retrieve a character at a certain position in the string.
Remark: The array functionality is broken in the early 1.1 development series of the compiler. It is expected
that this is fixed soon.

3.7.2 Variants in assignments and expressions

As can be seen from the definition above, most simple types can be assigned to a variant. Likewise,
a variant can be assigned to a simple type: If possible, the value of the variant will be converted to
the type that is being assigned to. This may fail: Assigning a variant containing a string to an integer
will fail unless the string represents a valid integer. In the following example, the first assignment
will work, the second will fail:

program testv3;

uses Variants;

39
 CHAPTER 3. TYPES

Var
V : Variant;
I : Integer;

begin
V:=’100’;
I:=V;
Writeln(’I : ’,I);
V:=’Something else’;
I:=V;
Writeln(’I : ’,I);
end.

The first assignment will work, but the second will not, as Something else cannot be converted
to a valid integer value. An EConvertError exception will be the result.
The result of an expression involving a variant will be of type variant again, but this can be assigned
to a variable of a different type - if the result can be converted to a variable of this type.
Note that expressions involving variants take more time to be evaluated, and should therefore be used
with caution. If a lot of calculations need to be made, it is best to avoid the use of variants.
When considering implicit type conversions (e.g. byte to integer, integer to double, char to string)
the compiler will ignore variants unless a variant appears explicitly in the expression.

3.7.3 Variants and interfaces

Remark: Dispatch interface support for variants is currently broken in the compiler.
Variants can contain a reference to an interface - a normal interface (descending from IInterface)
or a dispatchinterface (descending from IDispatch). Variants containing a reference to a dispatch
interface can be used to control the object behind it: the compiler will use late binding to perform
the call to the dispatch interface: there will be no run-time checking of the function names and
parameters or arguments given to the functions. The result type is also not checked. The compiler
will simply insert code to make the dispatch call and retrieve the result.
This means basically, that you can do the following on Windows:

Var
W : Variant;
V : String;

begin
W:=CreateOleObject(’[Link]’);
V:=[Link];
Writeln(’Installed version of MS Word is : ’,V);
end;

The line

V:=[Link];

is executed by inserting the necessary code to query the dispatch interface stored in the variant W, and
execute the call if the needed dispatch information is found.

40
Chapter 4

Variables

4.1 Definition
Variables are explicitly named memory locations with a certain type. When assigning values to
variables, the Free Pascal compiler generates machine code to move the value to the memory location
reserved for this variable. Where this variable is stored depends on where it is declared:

• Global variables are variables declared in a unit or program, but not inside a procedure or func-
tion. They are stored in fixed memory locations, and are available during the whole execution
time of the program.
• Local variables are declared inside a procedure or function. Their value is stored on the pro-
gram stack, i.e. not at fixed locations.

The Free Pascal compiler handles the allocation of these memory locations transparantly, although
this location can be influenced in the declaration.
The Free Pascal compiler also handles reading values from or writing values to the variables transparantly.
But even this can be explicitly handled by the programmer when using properties.
Variables must be explicitly declared when they are needed. No memory is allocated unless a variable
is declared. Using an variable identifier (for instance, a loop variable) which is not declared first, is
an error which will be reported by the compiler.

4.2 Declaration
The variables must be declared in a variable declaration section of a unit or a procedure or function.
It looks as follows:

Variable declaration

-
- variable declaration identifier : type -
= expression
- ; -
variable modifiers

41
 CHAPTER 4. VARIABLES

- variable modifiers
- absolute integer expression -
6 identifier
; export
; cvar
; external
string constant name string constant
- -

This means that the following are valid variable declarations:

Var
curterm1 : integer;

curterm2 : integer; cvar;

curterm3 : integer; cvar; external;

curterm4 : integer; external name ’curterm3’;

curterm5 : integer; external ’libc’ name ’curterm9’;

curterm6 : integer absolute curterm1;

curterm7 : integer; cvar; export;

curterm8 : integer; cvar; public;
curterm9 : integer; export name ’me’;
curterm10 : integer; public name ’ma’;

curterm11 : integer = 1 ;

The difference between these declarations is as follows:

1. The first form (curterm1) defines a regular variable. The compiler manages everything by
itself.
2. The second form (curterm2) declares also a regular variable, but specifies that the assembler
name for this variable equals the name of the variable as written in the source.
3. The third form (curterm3) declares a variable which is located externally: the compiler will
assume memory is located elsewhere, and that the assembler name of this location is specified
by the name of the variable, as written in the source. The name may not be specified.
4. The fourth form is completely equivalent to the third, it declares a variable which is stored
externally, and explicitly gives the assembler name of the location. If cvar is not used, the
name must be specified.
5. The fifth form is a variant of the fourth form, only the name of the library in which the memory
is reserved is specified as well.
6. The sixth form declares a variable (curterm6), and tells the compiler that it is stored in the
same location as another variable (curterm1)
7. The seventh form declares a variable (curterm7), and tells the compiler that the assembler
label of this variable should be the name of the variable (case sensitive) and must be made
public. (i.e. it can be referenced from other object files)
8. The eight form (curterm8) is equivalent to the seventh: ’public’ is an alias for ’export’.

42
 CHAPTER 4. VARIABLES

9. The ninth and tenth form are equivalent: they specify the assembler name of the variable.
10. the elevents form declares a variable (curterm11) and initializes it with a value (1 in the
above case).

Note that assembler names must be unique. It’s not possible to declare or export 2 variables with the
same assembler name.

4.3 Scope
Variables, just as any identifier, obey the general rules of scope. In addition, initialized variables are
initialized when they enter scope:

• Global initialized variables are initialized once, when the program starts.
• Local initialized variables are initialized each time the procedure is entered.

Note that the behaviour for local initialized variables is different from the one of a local typed con-
stant. A local typed constant behaves like a global initialized variable.

4.4 Thread Variables

For a program which uses threads, the variables can be really global, i.e. the same for all threads, or
thread-local: this means that each thread gets a copy of the variable. Local variables (defined inside
a procedure) are always thread-local. Global variables are normally the same for all threads. A
global variable can be declared thread-local by replacing the var keyword at the start of the variable
declaration block with Threadvar:

Threadvar
IOResult : Integer;

If no threads are used, the variable behaves as an ordinary variable. If threads are used then a copy is
made for each thread (including the main thread). Note that the copy is made with the original value
of the variable, not with the value of the variable at the time the thread is started.
Threadvars should be used sparingly: There is an overhead for retrieving or setting the variable’s
value. If possible at all, consider using local variables; they are always faster than thread variables.
Threads are not enabled by default. For more information about programming threads, see the chapter
on threads in the Programmers guide.

4.5 Properties
A global block can declare properties, just as they could be defined in a class. The difference is that
the global property does not need a class instance: there is only 1 instance of this property. Other
than that, a global property behaves like a class property. The read/write specifiers for the global
property must also be regular procedures, not methods. The concept of a global property is specific
to Free Pascal, and does not exist in Delphi.
The concept of a global property can be used to ’hide’ the location of the value, or to calculate the
value on the fly, or to check the values which are written to the property.
The declaration is as follows:

43
 CHAPTER 4. VARIABLES

Properties

-
- property definition identifier property specifiers -
property interface

-
- property interface : type identifier -
property parameter list
- -
index integerconstant

-
- property parameter list [ parameter declaration ] -
6 ;

- property specifiers
- -
read specifier write specifier default specifier

-
- read specifier read field or function -

-
- write specifier write field or procedure -

-
- default specifier default -
constant
nodefault

-
- field or procedure field identifier -
procedure identifier

-
- field or function field identifier -
function identifier

The following is an example:

{$mode objfpc}
unit testprop;

Interface

Function GetMyInt : Integer;

Procedure SetMyInt(Value : Integer);

Property
MyProp : Integer Read GetMyInt Write SetMyInt;

Implementation

Uses sysutils;

Var
FMyInt : Integer;

Function GetMyInt : Integer;

begin
Result:=FMyInt;

44
 CHAPTER 4. VARIABLES

end;

Procedure SetMyInt(Value : Integer);

begin
If ((Value mod 2)=1) then
Raise [Link](’MyProp can only contain even value’);
FMyInt:=Value;
end;

end.

The read/write specifiers can be hidden by declaring them in another unit which must be in the uses
clause of the unit. This can be used to hide the read/write access specifiers for programmers, just as
if they were in a private section of a class (discussed below). For the previous example, this could
look as follows:

{$mode objfpc}
unit testrw;

Interface

Function GetMyInt : Integer;

Procedure SetMyInt(Value : Integer);

Implementation

Uses sysutils;

Var
FMyInt : Integer;

Function GetMyInt : Integer;

begin
Result:=FMyInt;
end;

Procedure SetMyInt(Value : Integer);

begin
If ((Value mod 2)=1) then
Raise [Link](’Only even values are allowed’);
FMyInt:=Value;
end;

end.

The unit testprop would then look like:

{$mode objfpc}
unit testprop;

Interface

45
 CHAPTER 4. VARIABLES

uses testrw;

Property
MyProp : Integer Read GetMyInt Write SetMyInt;

Implementation

end.

46
Chapter 5

Objects

5.1 Declaration
Free Pascal supports object oriented programming. In fact, most of the compiler is written using
objects. Here we present some technical questions regarding object oriented programming in Free
Pascal. Objects should be treated as a special kind of record. The record contains all the fields that
are declared in the objects definition, and pointers to the methods that are associated to the objects’
type.
An object is declared just as a record would be declared; except that now,procedures and functions
can be declared as if they were part of the record. Objects can ”inherit” fields and methods from
”parent” objects. This means that these fields and methods can be used as if they were included in
the objects declared as a ”child” object.
Furthermore, a concept of visibility is introduced: fields, procedures and functions can be delcared
as public or private. By default, fields and methods are public, and are exported outside the
current unit. Fields or methods that are declared private are only accessible in the current unit.
The prototype declaration of an object is as follows:

object types

-
- object end -
packed heritage component list
object visibility specifier
6

-
- heritage ( object type identifier ) -

-
- component list -
field definition method definition
6 6

-
- field definition identifier list : type ; -

-
- method definition function header ; method directives -
procedure header
constructor header
desctuctor header

-
- method directives -
virtual ; abstract ; call modifiers ;

47
 CHAPTER 5. OBJECTS

-
- object visibility specifier private -
protected
public

As can be seen, as many private and public blocks as needed can be declared. Method
definitions are normal function or procedure declarations. Fields cannot be declared after meth-
ods in the same block, i.e. the following will generate an error when compiling:

Type MyObj = Object

Procedure Doit;
Field : Longint;
end;

But the following will be accepted:

Type MyObj = Object

Public
Procedure Doit;
Private
Field : Longint;
end;

because the field is in a different section.

Remark: Free Pascal also supports the packed object. This is the same as an object, only the elements (fields)
of the object are byte-aligned, just as in the packed record. The declaration of a packed object is
similar to the declaration of a packed record :

Type
TObj = packed object;
Constructor init;
...
end;
Pobj = ^TObj;
Var PP : Pobj;

Similarly, the {$PackRecords } directive acts on objects as well.

5.2 Fields
Object Fields are like record fields. They are accessed in the same way as a record field would be
accessed : by using a qualified identifier. Given the following declaration:

Type TAnObject = Object

AField : Longint;
Procedure AMethod;
end;
Var AnObject : TAnObject;

then the following would be a valid assignment:

[Link] := 0;

48
 CHAPTER 5. OBJECTS

Inside methods, fields can be accessed using the short identifier:

Procedure [Link];
begin
...
AField := 0;
...
end;

Or, one can use the self identifier. The self identifier refers to the current instance of the object:

Procedure [Link];
begin
...
[Link] := 0;
...
end;

One cannot access fields that are in a private section of an object from outside the objects’ methods. If
this is attempted anyway, the compiler will complain about an unknown identifier. It is also possible
to use the with statement with an object instance:

With AnObject do
begin
Afield := 12;
AMethod;
end;

In this example, between the begin and end, it is as if AnObject was prepended to the Afield
and Amethod identifiers. More about this in section 9.2.7, page 83

5.3 Constructors and destructors

As can be seen in the syntax diagram for an object declaration, Free Pascal supports constructors and
destructors. The programmer is responsible for calling the constructor and the destructor explicitly
when using objects. The declaration of a constructor or destructor is as follows:

Constructors and destructors

-
- constructor declaration constructor header ; subroutine block -

-
- destructor declaration destructor header ; subroutine block -

-
- constructor header constructor identifier -
qualified method identifier
- formal parameter list -

-
- destructor header destructor identifier -
qualified method identifier
- formal parameter list -

49
 CHAPTER 5. OBJECTS

A constructor/destructor pair is required if the object uses virtual methods. In the declaration of the
object type, a simple identifier should be used for the name of the constuctor or destructor. When
the constructor or destructor is implemented, A qualified method identifier should be used, i.e. an
identifier of the form [Link]. Free Pascal supports also the
extended syntax of the New and Dispose procedures. In case a dynamic variable of an object type
must be allocated the constructor’s name can be specified in the call to New. The New is implemented
as a function which returns a pointer to the instantiated object. Consider the following declarations:

Type
TObj = object;
Constructor init;
...
end;
Pobj = ^TObj;
Var PP : Pobj;

Then the following 3 calls are equivalent:

pp := new (Pobj,Init);

and

new(pp,init);

and also

new (pp);
pp^.init;

In the last case, the compiler will issue a warning that the extended syntax of new and dispose
must be used to generate instances of an object. It is possible to ignore this warning, but it’s better
programming practice to use the extended syntax to create instances of an object. Similarly, the
Dispose procedure accepts the name of a destructor. The destructor will then be called, before
removing the object from the heap.
In view of the compiler warning remark, the following chapter presents the Delphi approach to
object-oriented programming, and may be considered a more natural way of object-oriented pro-
gramming.

5.4 Methods
Object methods are just like ordinary procedures or functions, only they have an implicit extra pa-
rameter : self. Self points to the object with which the method was invoked. When implementing
methods, the fully qualified identifier must be given in the function header. When declaring methods,
a normal identifier must be given.

5.5 Method invocation

Methods are called just as normal procedures are called, only they have an object instance identifier
prepended to them (see also chapter 9, page 76). To determine which method is called, it is necessary
to know the type of the method. We treat the different types in what follows.

50
 CHAPTER 5. OBJECTS

Static methods

Static methods are methods that have been declared without a abstract or virtual keyword.
When calling a static method, the declared (i.e. compile time) method of the object is used. For
example, consider the following declarations:

Type
TParent = Object
...
procedure Doit;
...
end;
PParent = ^TParent;
TChild = Object(TParent)
...
procedure Doit;
...
end;
PChild = ^TChild;

As it is visible, both the parent and child objects have a method called Doit. Consider now the
following declarations and calls:

Var ParentA,ParentB : PParent;

Child : PChild;
ParentA := New(PParent,Init);
ParentB := New(PChild,Init);
Child := New(PChild,Init);
ParentA^.Doit;
ParentB^.Doit;
Child^.Doit;

Of the three invocations of Doit, only the last one will call [Link], the other two calls will
call [Link]. This is because for static methods, the compiler determines at compile time
which method should be called. Since ParentB is of type TParent, the compiler decides that
it must be called with [Link], even though it will be created as a TChild. There may
be times when the method that is actually called should depend on the actual type of the object at
run-time. If so, the method cannot be a static method, but must be a virtual method.

Virtual methods

To remedy the situation in the previous section, virtual methods are created. This is simply
done by appending the method declaration with the virtual modifier. Going back to the previous
example, consider the following alternative declaration:

Type
TParent = Object
...
procedure Doit;virtual;
...
end;
PParent = ^TParent;
TChild = Object(TParent)
...

51
 CHAPTER 5. OBJECTS

procedure Doit;virtual;
...
end;
PChild = ^TChild;

As it is visible, both the parent and child objects have a method called Doit. Consider now the
following declarations and calls :

Var ParentA,ParentB : PParent;

Child : PChild;
ParentA := New(PParent,Init);
ParentB := New(PChild,Init);
Child := New(PChild,Init);
ParentA^.Doit;
ParentB^.Doit;
Child^.Doit;

Now, different methods will be called, depending on the actual run-time type of the object. For
ParentA, nothing changes, since it is created as a TParent instance. For Child, the situation
also doesn’t change: it is again created as an instance of TChild. For ParentB however, the
situation does change: Even though it was declared as a TParent, it is created as an instance of
TChild. Now, when the program runs, before calling Doit, the program checks what the actual
type of ParentB is, and only then decides which method must be called. Seeing that ParentB is
of type TChild, [Link] will be called. The code for this run-time checking of the actual
type of an object is inserted by the compiler at compile time. The [Link] is said to override
the [Link]. It is possible to acces the [Link] from within the [Link],
with the inherited keyword:

Procedure [Link];
begin
inherited Doit;
...
end;

In the above example, when [Link] is called, the first thing it does is call [Link].
The inherited keyword cannot be used in static methods, only on virtual methods.

Abstract methods

An abstract method is a special kind of virtual method. A method can not be abstract if it is not virtual
(this is not obvious from the syntax diagram). An instance of an object that has an abstract method
cannot be created directly. The reason is obvious: there is no method where the compiler could jump
to ! A method that is declared abstract does not have an implementation for this method. It is up
to inherited objects to override and implement this method. Continuing our example, take a look at
this:

Type
TParent = Object
...
procedure Doit;virtual;abstract;
...
end;
PParent=^TParent;
TChild = Object(TParent)

52
 CHAPTER 5. OBJECTS

...
procedure Doit;virtual;
...
end;
PChild = ^TChild;

As it is visible, both the parent and child objects have a method called Doit. Consider now the
following declarations and calls :

Var ParentA,ParentB : PParent;

Child : PChild;
ParentA := New(PParent,Init);
ParentB := New(PChild,Init);
Child := New(PChild,Init);
ParentA^.Doit;
ParentB^.Doit;
Child^.Doit;

First of all, Line 3 will generate a compiler error, stating that one cannot generate instances of objects
with abstract methods: The compiler has detected that PParent points to an object which has an
abstract method. Commenting line 3 would allow compilation of the program.
Remark: If an abstract method is overridden, The parent method cannot be called with inherited, since
there is no parent method; The compiler will detect this, and complain about it, like this:

[Link](32,3) Error: Abstract methods can’t be called directly

If, through some mechanism, an abstract method is called at run-time, then a run-time error will
occur. (run-time error 211, to be precise)

5.6 Visibility
For objects, 3 visibility specifiers exist : private, protected and public. If a visibility speci-
fier is not specified, public is assumed. Both methods and fields can be hidden from a programmer
by putting them in a private section. The exact visibility rule is as follows:

Private All fields and methods that are in a private block, can only be accessed in the module
(i.e. unit or program) that contains the object definition. They can be accessed from inside the
object’s methods or from outside them e.g. from other objects’ methods, or global functions.
Protected Is the same as Private, except that the members of a Protected section are also
accessible to descendent types, even if they are implemented in other modules.
Public sections are always accessible, from everywhere. Fields and metods in a public section
behave as though they were part of an ordinary record type.

53
 Chapter 6

Classes

In the Delphi approach to Object Oriented Programming, everything revolves around the concept of
’Classes’. A class can be seen as a pointer to an object, or a pointer to a record.
Remark: In earlier versions of Free Pascal it was necessary, in order to use classes, to put the objpas unit in
the uses clause of a unit or program. This is no longer needed as of version 0.99.12. As of version
0.99.12 the system unit contains the basic definitions of TObject and TClass, as well as some
auxiliary methods for using classes. The objpas unit still exists, and contains some redefinitions of
basic types, so they coincide with Delphi types. The unit will be loaded automatically when the -S2
or -Sd options are specified.

6.1 Class definitions

The prototype declaration of a class is as follows :

Class types

-
- class end -
packed heritage component list
class visibility specifier
6

-
- heritage ( class type identifier ) -

-
- component list -
field definition method definition
6 6 property definition

-
- field definition identifier list : type ; -

-
- method definition function header ;-
class procedure header
constructor header
desctuctor header
- -
virtual ; call modifiers ;
; abstract
override
message integer constant
string constant

54
 CHAPTER 6. CLASSES

-
- class visibility specifier private -
protected
public
published

As many private, protected, published and public blocks as needed can be repeated.
Methods are normal function or procedure declarations. As can be seen, the declaration of a class is
almost identical to the declaration of an object. The real difference between objects and classes is
in the way they are created (see further in this chapter). The visibility of the different sections is as
follows:

Private All fields and methods that are in a private block, can only be accessed in the module
(i.e. unit) that contains the class definition. They can be accessed from inside the classes’
methods or from outside them (e.g. from other classes’ methods)
Protected Is the same as Private, except that the members of a Protected section are also
accessible to descendent types, even if they are implemented in other modules.
Public sections are always accessible.
Published Is the same as a Public section, but the compiler generates also type information that
is needed for automatic streaming of these classes. Fields defined in a published section
must be of class type. Array peroperties cannot be in a published section.

It is also possible to define class reference types:

Class reference type

-
- class of classtype -

Class reference types are used to create instances of a certain class, which is not yet known at compile
time, but which is specified at run time. Essentially, a variable of a class reference type contains a
pointer to the VMT of the speficied class. This can be used to construct an instance of the class
corresponding to the VMT. The following example shows how it works:

Type
TComponentClass = Class of TComponent;

Function CreateComponent(AClass : TComponentClass; AOwner : TComponent) : TComponent;

begin
// ...
Result:=[Link](AOwner);
// ...
end;

More about instantiating a class can be found in the next section.

55
 CHAPTER 6. CLASSES

6.2 Class instantiation

Classes must be created using their constructor. Remember that a class is a pointer to an object, so
when a variable of some class is declared, the compiler just allocates a pointer, not the entire object.
The constructor of a class returns a pointer to an initialized instance of the object. So, to initialize an
instance of some class, one would do the following :

ClassVar := [Link];

The extended syntax of new and dispose can be used to instantiate and destroy class instances.
That construct is reserved for use with objects only. Calling the constructor will provoke a call to
getmem, to allocate enough space to hold the class instance data. After that, the constuctor’s code
is executed. The constructor has a pointer to it’s data, in self.
Remark:

• The {$PackRecords } directive also affects classes. i.e. the alignment in memory of the
different fields depends on the value of the {$PackRecords } directive.
• Just as for objects and records, a packed class can be declared. This has the same effect as on
an object, or record, namely that the elements are aligned on 1-byte boundaries. i.e. as close
as possible.
• SizeOf(class) will return 4, since a class is but a pointer to an object. To get the size of
the class instance data, use the [Link] method.

6.3 Methods
6.3.1 invocation
Method invocation for classes is no different than for objects. The following is a valid method
invocation:

Var AnObject : TAnObject;

begin
AnObject := [Link];
[Link];

6.3.2 Virtual methods

Classes have virtual methods, just as objects do. There is however a difference between the two.
For objects, it is sufficient to redeclare the same method in a descendent object with the keyword
virtual to override it. For classes, the situation is different: virtual methods must be overridden
with the override keyword. Failing to do so, will start a new batch of virtual methods, hiding the
previous one. The Inherited keyword will not jump to the inherited method, if virtual was used.
The following code is wrong:

Type
ObjParent = Class
Procedure MyProc; virtual;
end;
ObjChild = Class(ObjPArent)
Procedure MyProc; virtual;
end;

56
 CHAPTER 6. CLASSES

The compiler will produce a warning:

Warning: An inherited method is hidden by [Link]

The compiler will compile it, but using Inherited can produce strange effects.
The correct declaration is as follows:

Type ObjParent = Class

Procedure MyProc; virtual;
end;
ObjChild = Class(ObjPArent)
Procedure MyProc; override;
end;

This will compile and run without warnings or errors.

6.3.3 Class methods

Class methods are methods that do not have an instance, but which follow the scoping and inheritance
rules of a class. They can be called from inside a regular method, but can also be called using a class
identifier:

Var
AClass : TClass;

begin
..
if CompareText([Link],’TCOMPONENT’)=0 then
...

But calling them from an instance is also possible:

Var
MyClass : TObject;

begin
..
if [Link](’TCOMPONENT’) then
...

Inside a class method, the <var>self</var> identifier points to the VMT table of the class. No fields,
properties or regular methods are available inside a class method. Accessing a regular property or
method will result in a compiler error. The reverse is possible: a class method can be called from a
regular method.
Note that class methods can be virtual, and can be overridden.
Class methods cannot be used as read or write specifiers for a property.

6.3.4 Message methods

New in classes are message methods. Pointers to message methods are stored in a special table,
together with the integer or string cnstant that they were declared with. They are primarily intended to

57
 CHAPTER 6. CLASSES

ease programming of callback functions in several GUI toolkits, such as Win32 or GTK. In difference
with Delphi, Free Pascal also accepts strings as message identifiers.
Message methods that are declared with an integer constant can take only one var argument (typed
or not):

Procedure [Link](Var Msg); Message 1;

The method implementation of a message function is no different from an ordinary method. It is also
possible to call a message method directly, but this should not be done. Instead, the [Link]
method should be used.
The [Link] method can be used to call a message handler. It is declared in the
system unit and will accept a var parameter which must have at the first position a cardinal with the
message ID that should be called. For example:

Type
TMsg = Record
MSGID : Cardinal
Data : Pointer;
Var
Msg : TMSg;

[Link] (Msg);

In this example, the Dispatch method will look at the object and all it’s ancestors (starting at the
object, and searching up the class tree), to see if a message method with message MSGID has been
declared. If such a method is found, it is called, and passed the Msg parameter.
If no such method is found, DefaultHandler is called. DefaultHandler is a virtual method
of TObject that doesn’t do anything, but which can be overridden to provide any processing that
might be needed. DefaultHandler is declared as follows:

procedure defaulthandler(var message);virtual;

In addition to the message method with a Integer identifier, Free Pascal also supports a message
method with a string identifier:

Procedure [Link](Var Msg); Message ’OnClick’;

The working of the string message handler is the same as the ordinary integer message handler:
The [Link] method can be used to call a message handler. It is declared in
the system unit and will accept one parameter which must have at the first position a string with the
message ID that should be called. For example:

Type
TMsg = Record
MsgStr : String[10]; // Arbitrary length up to 255 characters.
Data : Pointer;
Var
Msg : TMSg;

[Link] (Msg);

In this example, the DispatchStr method will look at the object and all it’s ancestors (starting at
the object, and searching up the class tree), to see if a message method with message MsgStr has
been declared. If such a method is found, it is called, and passed the Msg parameter.

58
 CHAPTER 6. CLASSES

If no such method is found, DefaultHandlerStr is called. DefaultHandlerStr is a virtual

method of TObject that doesn’t do anything, but which can be overridden to provide any processing
that might be needed. DefaultHandlerStr is declared as follows:

procedure DefaultHandlerStr(var message);virtual;

In addition to this mechanism, a string message method accepts a self parameter:

[Link](Data : Pointer; Self : TMyObject);Message ’OnClick’;

When encountering such a method, the compiler will generate code that loads the Self parameter
into the object instance pointer. The result of this is that it is possible to pass Self as a parameter to
such a method.
Remark: The type of the Self parameter must be of the same class as the class the method is defined in.

6.4 Properties
Classes can contain properties as part of their fields list. A property acts like a normal field, i.e. its
value can be retrieved or set, but it allows to redirect the access of the field through functions and
procedures. They provide a means to associate an action with an assignment of or a reading from a
class ’field’. This allows for e.g. checking that a value is valid when assigning, or, when reading,
it allows to construct the value on the fly. Moreover, properties can be read-only or write only. The
prototype declaration of a property is as follows:

Properties

-
- property definition property identifier -
property interface
- property specifiers -

-
- property interface : type identifier -
property parameter list
- -
index integerconstant

-
- property parameter list [ parameter declaration ] -
6 ;

- property specifiers
- -
read specifier write specifier default specifier

-
- read specifier read field or method -

-
- write specifier write field or method -

-
- default specifier default -
constant
nodefault

-
- field or method field identifier -
method identifier

59
 CHAPTER 6. CLASSES

A read specifier is either the name of a field that contains the property, or the name of a
method function that has the same return type as the property type. In the case of a simple type, this
function must not accept an argument. A read specifier is optional, making the property write-
only. Note that class methods cannot be used as read specifiers. A write specifier is optional:
If there is no write specifier, the property is read-only. A write specifier is either the name of
a field, or the name of a method procedure that accepts as a sole argument a variable of the same type
as the property. The section (private, published) in which the specified function or procedure
resides is irrelevant. Usually, however, this will be a protected or private method. Example: Given
the following declaration:

Type
MyClass = Class
Private
Field1 : Longint;
Field2 : Longint;
Field3 : Longint;
Procedure Sety (value : Longint);
Function Gety : Longint;
Function Getz : Longint;
Public
Property X : Longint Read Field1 write Field2;
Property Y : Longint Read GetY Write Sety;
Property Z : Longint Read GetZ;
end;
Var MyClass : TMyClass;

The following are valid statements:

WriteLn (’X : ’,MyClass.X);

WriteLn (’Y : ’,MyClass.Y);
WriteLn (’Z : ’,MyClass.Z);
MyClass.X := 0;
MyClass.Y := 0;

But the following would generate an error:

MyClass.Z := 0;

because Z is a read-only property. What happens in the above statements is that when a value needs
to be read, the compiler inserts a call to the various getNNN methods of the object, and the result of
this call is used. When an assignment is made, the compiler passes the value that must be assigned
as a paramater to the various setNNN methods. Because of this mechanism, properties cannot be
passed as var arguments to a function or procedure, since there is no known address of the property
(at least, not always). If the property definition contains an index, then the read and write specifiers
must be a function and a procedure. Moreover, these functions require an additional parameter : An
integer parameter. This allows to read or write several properties with the same function. For this,
the properties must have the same type. The following is an example of a property with an index:

{$mode objfpc}
Type TPoint = Class(TObject)
Private
FX,FY : Longint;
Function GetCoord (Index : Integer): Longint;
Procedure SetCoord (Index : Integer; Value : longint);

60
 CHAPTER 6. CLASSES

Public
Property X : Longint index 1 read GetCoord Write SetCoord;
Property Y : Longint index 2 read GetCoord Write SetCoord;
Property Coords[Index : Integer]:Longint Read GetCoord;
end;
Procedure [Link] (Index : Integer; Value : Longint);
begin
Case Index of
1 : FX := Value;
2 : FY := Value;
end;
end;
Function [Link] (INdex : Integer) : Longint;
begin
Case Index of
1 : Result := FX;
2 : Result := FY;
end;
end;
Var P : TPoint;
begin
P := [Link];
P.X := 2;
P.Y := 3;
With P do
WriteLn (’X=’,X,’ Y=’,Y);
end.

When the compiler encounters an assignment to X, then SetCoord is called with as first parameter
the index (1 in the above case) and with as a second parameter the value to be set. Conversely, when
reading the value of X, the compiler calls GetCoord and passes it index 1. Indexes can only be
integer values. Array propertie also exist. These are properties that accept an index, just as an array
does. Only now the index doesn’t have to be an ordinal type, but can be any type.
A read specifier for an array property is the name method function that has the same return
type as the property type. The function must accept as a sole arguent a variable of the same type as
the index type. For an array property, one cannot specify fields as read specifiers.
A write specifier for an array property is the name of a method procedure that accepts two
arguments: The first argument has the same type as the index, and the second argument is a parameter
of the same type as the property type. As an example, see the following declaration:

Type TIntList = Class

Private
Function GetInt (I : Longint) : longint;
Function GetAsString (A : String) : String;
Procedure SetInt (I : Longint; Value : Longint;);
Procedure SetAsString (A : String; Value : String);
Public
Property Items [i : Longint] : Longint Read GetInt
Write SetInt;
Property StrItems [S : String] : String Read GetAsString
Write SetAsstring;
end;
Var AIntList : TIntList;

61
 CHAPTER 6. CLASSES

Then the following statements would be valid:

[Link][26] := 1;
[Link][’twenty-five’] := ’zero’;
WriteLn (’Item 26 : ’,[Link][26]);
WriteLn (’Item 25 : ’,[Link][’twenty-five’]);

While the following statements would generate errors:

[Link][’twenty-five’] := 1;
[Link][26] := ’zero’;

Because the index types are wrong. Array properties can be declared as default properties. This
means that it is not necessary to specify the property name when assigning or reading it. If, in the
previous example, the definition of the items property would have been

Property Items[i : Longint]: Longint Read GetInt

Write SetInt; Default;

Then the assignment

[Link][26] := 1;

Would be equivalent to the following abbreviation.

AIntList[26] := 1;

Only one default property per class is allowed, and descendent classes cannot redeclare the default
property.

62
Chapter 7

Interfaces

7.1 Definition
As of version 1.1, FPC supports interfaces. Interfaces are an alternative to multiple inheritance
(where a class can have multiple parent classes) as implemented for instance in C++. An interface
is basically a named set of methods and properties: A class that implements the interface provides
all the methods as they are enumerated in the Interface definition. It is not possible for a class to
implement only part of the interface: it is all or nothing.
Interfaces can also be ordered in a hierarchy, exactly as classes: An interface definition that inherits
from another interface definition contains all the methods from the parent interface, as well as the
methods explicitly named in the interface definition. A class implementing an interface must then
implement all members of the interface as well as the methods of the parent interface(s).
An interface can be uniquely identified by a GUID (GUID is an acronym for Globally Unique Iden-
tifier, a 128-bit integer guaranteed always to be unique1 . Especially on Windows systems, the GUID
of an interface can and must be used when using COM.
The definition of an Interface has the following form:

Interface type

-
- Interface end -
heritage [’ GUID ’] component list

-
- heritage ( interface type identifier ) -

-
- component list method definition -
6 property definition

Along with this definition the following must be noted:

• Interfaces can only be used in DELPHI mode or in OBJFPC mode.

• There are no visibility specifiers. All members are public (indeed, it would make little sense
to make them private or protected).
1 In theory, of course.

63
 CHAPTER 7. INTERFACES

• The properties declared in an interface can only have methods as read and write specifiers.
• There are no constructors or destructors. Instances of interfaces cannot be created directly:
instead, an instance of a class implementing the interface must be created.
• Only calling convention modifiers may be present in the definition of a method. Modifiers
as virtual, abstract or dynamic, and hence also override cannot be present in the
definition of a interface definition.

7.2 Interface identification: A GUID

An interface can be identified by a GUID. This is a 128-bit number, which is represented in a text
representation (a string literal):

[’{HHHHHHHH-HHHH-HHHH-HHHH-HHHHHHHHHHHH}’]

Each H character represents a hexadecimal number (0-9,A-F). The format contains 8-4-4-4-12 num-
bers. A GUID can also be represented by the following record, defined in the objpas unit (included
automatically when in DELPHI or OBJFPC mode:

PGuid = ^TGuid;
TGuid = packed record
case integer of
1 : (
Data1 : DWord;
Data2 : word;
Data3 : word;
Data4 : array[0..7] of byte;
);
2 : (
D1 : DWord;
D2 : word;
D3 : word;
D4 : array[0..7] of byte;
);
end;

A constant of type TGUID can be specified using a string literal:

{$mode objfpc}
program testuid;

Const
MyGUID : TGUID = ’{10101010-1010-0101-1001-110110110110}’;

begin
end.

Normally, the GUIDs are only used in Windows, when using COM interfaces. More on this in the
next section.

64
 CHAPTER 7. INTERFACES

7.3 Interfaces and COM

When using interfaces on Windows which should be available to the COM subsystem, the calling
convention should be stdcall - this is not the default Free Pascal calling convention, so it should
be specified explicitly.
COM does not know properties. It only knows methods. So when specifying property definitions
as part of an interface definition, be aware that the properties will only be known in the Free Pascal
compiled program: other Windows programs will not be aware of the property definitions. For this
reason, property definitions must always have interface methods as the read/write specifiers.

Interface implementations
When a class implements an interface, it should implement all methods of the interface. If a method
of an interface is not implemented, then the compiler will give an error. For example:

Type
IMyInterface = Interface
Function MyFunc : Integer;
Function MySecondFunc : Integer;
end;

TMyClass = Class(TInterfacedObject,IMyInterface)
Function MyFunc : Integer;
Function MyOtherFunc : Integer;
end;

Function [Link] : Integer;

begin
Result:=23;
end;

Function [Link] : Integer;

begin
Result:=24;
end;

will result in a compiler error:

Error: No matching implementation for interface method

"[Link]:LongInt" found

At the moment of writing, the compiler does not yet support providing aliases for an interface as in
Delphi. i.e. the following will not yet compile:

ype
IMyInterface = Interface
Function MyFunc : Integer;
end;

TMyClass = Class(TInterfacedObject,IMyInterface)

65
 CHAPTER 7. INTERFACES

Function MyOtherFunction : Integer;

// The following fails in FPC.
Function [Link] = MyOtherFunction;
end;

This declaration should tell the compiler that the MyFunc method of the IMyInterface interface
is implemented in the MyOtherFunction method of the TMyClass class.

7.4 CORBA and other Interfaces

COM is not the only architecture where interfaces are used. CORBA knows interfaces, UNO (the
OpenOffice API) uses interfaces, and Java as well. These languages do not know the IUnknown
interface used as the basis of all interfaces in COM. It would therefore be a bad idea if an inter-
face automatically descended from IUnknown if no parent interface was specified. Therefore, a
directive {$INTERFACES} was introduced in Free Pascal: it specifies what the parent interface is
of an interface, declared without parent. More information about this directive can be found in the
Programmers guide.
Note that COM interfaces are by default reference counted. CORBA interfaces are not necessarily
reference counted.

66
 Chapter 8

Expressions

Expressions occur in assignments or in tests. Expressions produce a value, of a certain type. Expres-
sions are built with two components: Operators and their operands. Usually an operator is binary, i.e.
it requires 2 operands. Binary operators occur always between the operands (as in X/Y). Sometimes
an operator is unary, i.e. it requires only one argument. A unary operator occurs always before the
operand, as in -X.
When using multiple operands in an expression, the precedence rules of table (8.1) are used. When

Table 8.1: Precedence of operators

Operator Precedence Category

Not, @ Highest (first) Unary operators
* / div mod and shl shr as Second Multiplying operators
+ - or xor Third Adding operators
< <> < > <= >= in is Lowest (Last) relational operators

determining the precedence, the compiler uses the following rules:

1. In operations with unequal precedences the operands belong to the operater with the high-
est precedence. For example, in 5*3+7, the multiplication is higher in precedence than the
addition, so it is executed first. The result would be 22.
2. If parentheses are used in an expression, their contents is evaluated first. Thus, 5*(3+7)
would result in 50.

Remark: The order in which expressions of the same precedence are evaluated is not guaranteed to be left-
to-right. In general, no assumptions on which expression is evaluated first should be made in such a
case. The compiler will decide which expression to evaluate first based on optimization rules. Thus,
in the following expression:

a := g(3) + f(2);

f(2) may be executed before g(3). This behaviour is distinctly different from Delphior Turbo
Pascal.
If one expression must be executed before the other, it is necessary to split up the statement using
temporary results:

67
 CHAPTER 8. EXPRESSIONS

e1 := g(3);
a := e1 + f(2);

8.1 Expression syntax

An expression applies relational operators to simple expressions. Simple expressions are a series of
terms (what a term is, is explained below), joined by adding operators.

Expressions

-
- expression simple expression -
* simple expression
<=
>
>=
=
<>
in
is

-
- simple expression term -
6 +
-
or
xor

The following are valid expressions:

GraphResult<>grError
(DoItToday=Yes) and (DoItTomorrow=No);
Day in Weekend

And here are some simple expressions:

A + B
-Pi
ToBe or NotToBe

Terms consist of factors, connected by multiplication operators.

Terms

-
- term factor -
6 *
/
div
mod
and
shl
shr
as

68
 CHAPTER 8. EXPRESSIONS

Here are some valid terms:

2 * Pi
A Div B
(DoItToday=Yes) and (DoItTomorrow=No);

Factors are all other constructions:

Factors

-
- factor ( expression ) -
variable reference
function call
unsigned constant
not factor
sign factor
set constructor
value typecast
address factor

-
- unsigned constant unsigned number -
character string
constant identifier
Nil

8.2 Function calls

Function calls are part of expressions (although, using extended syntax, they can be statements too).
They are constructed as follows:

Function calls

-
- function call function identifier -
method designator actual parameter list
qualified method designator
variable reference

-
- actual parameter list ( ) -
expression
6 ,

The variable reference must be a procedural type variable reference. A method designator can
only be used inside the method of an object. A qualified method designator can be used outside
object methods too. The function that will get called is the function with a declared parameter list
that matches the actual parameter list. This means that

1. The number of actual parameters must equal the number of declared parameters (unless default
parameter values are used).

69
 CHAPTER 8. EXPRESSIONS

2. The types of the parameters must be compatible. For variable reference parameters, the pa-
rameter types must be exactly the same.

If no matching function is found, then the compiler will generate an error. Depending on the fact of
the function is overloaded (i.e. multiple functions with the same name, but different parameter lists)
the error will be different. There are cases when the compiler will not execute the function call in an
expression. This is the case when assigning a value to a procedural type variable, as in the following
example:

Type
FuncType = Function: Integer;
Var A : Integer;
Function AddOne : Integer;
begin
A := A+1;
AddOne := A;
end;
Var F : FuncType;
N : Integer;
begin
A := 0;
F := AddOne; { Assign AddOne to F, Don’t call AddOne}
N := AddOne; { N := 1 !!}
end.

In the above listing, the assigment to F will not cause the function AddOne to be called. The assign-
ment to N, however, will call AddOne. A problem with this syntax is the following construction:

If F = AddOne Then
DoSomethingHorrible;

Should the compiler compare the addresses of F and AddOne, or should it call both functions, and
compare the result ? Free Pascal solves this by deciding that a procedural variable is equivalent to a
pointer. Thus the compiler will give a type mismatch error, since AddOne is considered a call to a
function with integer result, and F is a pointer, Hence a type mismatch occurs. How then, should one
compare whether F points to the function AddOne ? To do this, one should use the address operator
@:

If F = @AddOne Then
WriteLn (’Functions are equal’);

The left hand side of the boolean expression is an address. The right hand side also, and so the
compiler compares 2 addresses. How to compare the values that both functions return ? By adding
an empty parameter list:

If F()=Addone then
WriteLn (’Functions return same values ’);

Remark that this behaviour is not compatible with Delphi syntax.

8.3 Set constructors

When a set-type constant must be entered in an expression, a set constructor must be given. In
essence this is the same thing as when a type is defined, only there is no identifier to identify the set
with. A set constructor is a comma separated list of expressions, enclosed in square brackets.

70
 CHAPTER 8. EXPRESSIONS

Set constructors

-
- set constructor [ ] -
set group
6 ,

-
- set group expression -
.. expression

All set groups and set elements must be of the same ordinal type. The empty set is denoted by [],
and it can be assigned to any type of set. A set group with a range [A..Z] makes all values in the
range a set element. If the first range specifier has a bigger ordinal value than the second the set is
empty, e.g., [Z..A] denotes an empty set. The following are valid set constructors:

[today,tomorrow]
[Monday..Friday,Sunday]
[ 2, 3*2, 6*2, 9*2 ]
[’A’..’Z’,’a’..’z’,’0’..’9’]

8.4 Value typecasts

Sometimes it is necessary to change the type of an expression, or a part of the expression, to be able
to be assignment compatible. This is done through a value typecast. The syntax diagram for a value
typecast is as follows:

Typecasts

-
- value typecast type identifier ( expression ) -

Value typecasts cannot be used on the left side of assignments, as variable typecasts. Here are some
valid typecasts:

Byte(’A’)
Char(48)
boolean(1)
longint(@Buffer)

The type size of the expression and the size of the type cast must be the same. That is, the following
doesn’t work:

Integer(’A’)
Char(4875)
boolean(100)
Word(@Buffer)

This is different from Delphi or Turbo Pascal behaviour.

71
 CHAPTER 8. EXPRESSIONS

8.5 The @ operator

The address operator @ returns the address of a variable, procedure or function. It is used as follows:

Address factor

-
- addressfactor @ variable reference -
procedure identifier
function identifier
qualified method identifier

The @ operator returns a typed pointer if the $T switch is on. If the $T switch is off then the address
operator returns an untyped pointer, which is assigment compatible with all pointer types. The type
of the pointer is ˆT, where T is the type of the variable reference. For example, the following will
compile

Program tcast;
{$T-} { @ returns untyped pointer }

Type art = Array[1..100] of byte;

Var Buffer : longint;
PLargeBuffer : ^art;

begin
PLargeBuffer := @Buffer;
end.

Changing the {$T-} to {$T+} will prevent the compiler from compiling this. It will give a type
mismatch error. By default, the address operator returns an untyped pointer. Applying the address
operator to a function, method, or procedure identifier will give a pointer to the entry point of that
function. The result is an untyped pointer. By default, the address operator must be used if a value
must be assigned to a procedural type variable. This behaviour can be avoided by using the -So or
-S2 switches, which result in a more compatible Delphi or Turbo Pascal syntax.

8.6 Operators
Operators can be classified according to the type of expression they operate on. We will discuss them
type by type.

8.6.1 Arithmetic operators

Arithmetic operators occur in arithmetic operations, i.e. in expressions that contain integers or reals.
There are 2 kinds of operators : Binary and unary arithmetic operators. Binary operators are listed
in table (8.2), unary operators are listed in table (8.3). With the exception of Div and Mod, which
accept only integer expressions as operands, all operators accept real and integer expressions as
operands. For binary operators, the result type will be integer if both operands are integer type
expressions. If one of the operands is a real type expression, then the result is real. As an exception :
division (/) results always in real values. For unary operators, the result type is always equal to the
expression type. The division (/) and Mod operator will cause run-time errors if the second argument
is zero. The sign of the result of a Mod operator is the same as the sign of the left side operand of the
Mod operator. In fact, the Mod operator is equivalent to the following operation :

72
 CHAPTER 8. EXPRESSIONS

Table 8.2: Binary arithmetic operators

Operator Operation
+ Addition
- Subtraction
* Multiplication
/ Division
Div Integer division
Mod Remainder

Table 8.3: Unary arithmetic operators

Operator Operation
+ Sign identity
- Sign inversion

I mod J = I - (I div J) * J

but it executes faster than the right hand side expression.

8.6.2 Logical operators

Logical operators act on the individual bits of ordinal expressions. Logical operators require operands
that are of an integer type, and produce an integer type result. The possible logical operators are listed
in table (8.4). The following are valid logical expressions:

Table 8.4: Logical operators

Operator Operation
not Bitwise negation (unary)
and Bitwise and
or Bitwise or
xor Bitwise xor
shl Bitwise shift to the left
shr Bitwise shift to the right

A shr 1 { same as A div 2, but faster}

Not 1 { equals -2 }
Not 0 { equals -1 }
Not -1 { equals 0 }
B shl 2 { same as B * 4 for integers }
1 or 2 { equals 3 }
3 xor 1 { equals 2 }

73
 CHAPTER 8. EXPRESSIONS

8.6.3 Boolean operators

Boolean operators can be considered logical operations on a type with 1 bit size. Therefore the shl
and shr operations have little sense. Boolean operators can only have boolean type operands, and
the resulting type is always boolean. The possible operators are listed in table (8.5)

Table 8.5: Boolean operators

Operator Operation
not logical negation (unary)
and logical and
or logical or
xor logical xor

Remark: Boolean expressions are always evaluated with short-circuit evaluation. This means that from the
moment the result of the complete expression is known, evaluation is stopped and the result is re-
turned. For instance, in the following expression:

B := True or MaybeTrue;

The compiler will never look at the value of MaybeTrue, since it is obvious that the expression will
always be true. As a result of this strategy, if MaybeTrue is a function, it will not get called ! (This
can have surprising effects when used in conjunction with properties)

8.6.4 String operators

There is only one string operator : +. It’s action is to concatenate the contents of the two strings
(or characters) it stands between. One cannot use + to concatenate null-terminated (PChar) strings.
The following are valid string operations:

’This is ’ + ’VERY ’ + ’easy !’

Dirname+’\’

The following is not:

Var Dirname = Pchar;

...
Dirname := Dirname+’\’;

Because Dirname is a null-terminated string.

8.6.5 Set operators

The following operations on sets can be performed with operators: Union, difference and intersec-
tion. The operators needed for this are listed in table (8.6). The set type of the operands must be the
same, or an error will be generated by the compiler.

8.6.6 Relational operators

The relational operators are listed in table (8.7) Left and right operands must be of the same type.

74
 CHAPTER 8. EXPRESSIONS

Table 8.6: Set operators

Operator Action
+ Union
- Difference
* Intersection

Table 8.7: Relational operators

Operator Action
= Equal
<> Not equal
< Stricty less than
> Strictly greater than
<= Less than or equal
>= Greater than or equal
in Element of

Only integer and real types can be mixed in relational expressions. Comparing strings is done on
the basis of their ASCII code representation. When comparing pointers, the addresses to which they
point are compared. This also is true for PChar type pointers. To compare the strings the Pchar
point to, the StrComp function from the strings unit must be used. The in returns True if the
left operand (which must have the same ordinal type as the set type, and which must be in the range
0..255) is an element of the set which is the right operand, otherwise it returns False

75
Chapter 9

Statements

The heart of each algorithm are the actions it takes. These actions are contained in the statements of
a program or unit. Each statement can be labeled and jumped to (within certain limits) with Goto
statements. This can be seen in the following syntax diagram:

Statements

-
- statement -
label : simple statement
structured statement
asm statement

A label can be an identifier or an integer digit.

9.1 Simple statements

A simple statement cannot be decomposed in separate statements. There are basically 4 kinds of
simple statements:

Simple statements

-
- simple statement assignment statement -
procedure statement
goto statement
raise statement

Of these statements, the raise statement will be explained in the chapter on Exceptions (chapter 13,
page 113)

9.1.1 Assignments
Assignments give a value to a variable, replacing any previous value the variable might have had:

76
 CHAPTER 9. STATEMENTS

Assignments

-
- assignment statement variable reference := expression -
function identifier +=
-=
*=
/=

In addition to the standard Pascal assignment operator ( := ), which simply replaces the value of
the varable with the value resulting from the expression on the right of the := operator, Free Pascal
supports some c-style constructions. All available constructs are listed in table (9.1). For these

Table 9.1: Allowed C constructs in Free Pascal

Assignment Result
a += b Adds b to a, and stores the result in a.
a -= b Substracts b from a, and stores the result in a.
a *= b Multiplies a with b, and stores the result in a.
a /= b Divides a through b, and stores the result in a.

constructs to work, the -Sc command-line switch must be specified.

Remark: These constructions are just for typing convenience, they don’t generate different code. Here are
some examples of valid assignment statements:

X := X+Y;
X+=Y; { Same as X := X+Y, needs -Sc command line switch}
X/=2; { Same as X := X/2, needs -Sc command line switch}
Done := False;
Weather := Good;
MyPi := 4* Tan(1);

9.1.2 Procedure statements

Procedure statements are calls to subroutines. There are different possibilities for procedure calls: A
normal procedure call, an object method call (fully qualified or not), or even a call to a procedural
type variable. All types are present in the following diagram.

Procedure statements

-
- procedure statement procedure identifier -
method identifier actual parameter list
qualified method identifier
variable reference

The Free Pascal compiler will look for a procedure with the same name as given in the procedure
statement, and with a declared parameter list that matches the actual parameter list. The following
are valid procedure statements:

77
 CHAPTER 9. STATEMENTS

Usage;
WriteLn(’Pascal is an easy language !’);
Doit();

9.1.3 Goto statements

Free Pascal supports the goto jump statement. Its prototype syntax is

Goto statement

-
- goto statement goto label -

When using goto statements, the following must be kept in mind:

1. The jump label must be defined in the same block as the Goto statement.
2. Jumping from outside a loop to the inside of a loop or vice versa can have strange effects.
3. To be able to use the Goto statement, the -Sg compiler switch must be used.

Goto statements are considered bad practice and should be avoided as much as possible. It is always
possible to replace a goto statement by a construction that doesn’t need a goto, although this
construction may not be as clear as a goto statement. For instance, the following is an allowed goto
statement:

label
jumpto;
...
Jumpto :
Statement;
...
Goto jumpto;
...

9.2 Structured statements

Structured statements can be broken into smaller simple statements, which should be executed re-
peatedly, conditionally or sequentially:

Structured statements

-
- structured statement compound statement -
repetitive statement
conditional statement
exception statement
with statement

Conditional statements come in 2 flavours :

78
 CHAPTER 9. STATEMENTS

Conditional statements

-
- conditional statement if statement -
case statement

Repetitive statements come in 3 flavours:

Repetitive statements

-
- repetitive statement for statament -
repeat statement
while statement

The following sections deal with each of these statements.

9.2.1 Compound statements

Compound statements are a group of statements, separated by semicolons, that are surrounded by
the keywords Begin and End. The Last statement doesn’t need to be followed by a semicolon,
although it is allowed. A compound statement is a way of grouping statements together, executing
the statements sequentially. They are treated as one statement in cases where Pascal syntax expects
1 statement, such as in if ... then statements.

Compound statements

-
- compound statement begin statement end -
6 ;

9.2.2 The Case statement

Free Pascal supports the case statement. Its syntax diagram is

Case statement

-
- case statement case expression of case end -
6; else part ;

-
- case constant : statement -
6 .. constant
,

-
- else part else statement -

79
 CHAPTER 9. STATEMENTS

The constants appearing in the various case parts must be known at compile-time, and can be of the
following types : enumeration types, Ordinal types (except boolean), and chars. The expression must
be also of this type, or a compiler error will occur. All case constants must have the same type. The
compiler will evaluate the expression. If one of the case constants values matches the value of the
expression, the statement that follows this constant is executed. After that, the program continues
after the final end. If none of the case constants match the expression value, the statement after
the else keyword is executed. This can be an empty statement. If no else part is present, and no
case constant matches the expression value, program flow continues after the final end. The case
statements can be compound statements (i.e. a begin..End block).
Remark: Contrary to Turbo Pascal, duplicate case labels are not allowed in Free Pascal, so the following code
will generate an error when compiling:

Var i : integer;
...
Case i of
3 : DoSomething;
1..5 : DoSomethingElse;
end;

The compiler will generate a Duplicate case label error when compiling this, because the 3
also appears (implicitly) in the range 1..5. This is similar to Delphi syntax.
The following are valid case statements:

Case C of
’a’ : WriteLn (’A pressed’);
’b’ : WriteLn (’B pressed’);
’c’ : WriteLn (’C pressed’);
else
WriteLn (’unknown letter pressed : ’,C);
end;

Or

Case C of
’a’,’e’,’i’,’o’,’u’ : WriteLn (’vowel pressed’);
’y’ : WriteLn (’This one depends on the language’);
else
WriteLn (’Consonant pressed’);
end;

Case Number of
1..10 : WriteLn (’Small number’);
11..100 : WriteLn (’Normal, medium number’);
else
WriteLn (’HUGE number’);
end;

9.2.3 The If..then..else statement

The If .. then .. else.. prototype syntax is

If then statements

80
 CHAPTER 9. STATEMENTS

-
- if statement if expression then statement -
else statement

The expression between the if and then keywords must have a boolean return type. If the expres-
sion evaluates to True then the statement following then is executed.
If the expression evaluates to False, then the statement following else is executed, if it is present.
Be aware of the fact that the boolean expression will be short-cut evaluated. (Meaning that the
evaluation will be stopped at the point where the outcome is known with certainty) Also, before
the else keyword, no semicolon (;) is allowed, but all statements can be compound statements.
In nested If.. then .. else constructs, some ambiguity may araise as to which else
statement pairs with which if statement. The rule is that the else keyword matches the first if
keyword not already matched by an else keyword. For example:

If exp1 Then
If exp2 then
Stat1
else
stat2;

Despite it’s appearance, the statement is syntactically equivalent to

If exp1 Then
begin
If exp2 then
Stat1
else
stat2
end;

and not to

{ NOT EQUIVALENT }
If exp1 Then
begin
If exp2 then
Stat1
end
else
stat2

If it is this latter construct is needed, the begin and end keywords must be present. When in doubt,
it is better to add them.
The following is a valid statement:

If Today in [Monday..Friday] then

WriteLn (’Must work harder’)
else
WriteLn (’Take a day off.’);

9.2.4 The For..to/downto..do statement

Free Pascal supports the For loop construction. A for loop is used in case one wants to calculated
something a fixed number of times. The prototype syntax is as follows:

81
 CHAPTER 9. STATEMENTS

For statement

-
- for statement for control variable := initial value to -
downto
- final value do statement -

-
- control variable variable identifier -

-
- initial value expression -

-
- final value expression -

Statement can be a compound statement. When this statement is encountered, the control variable
is initialized with the initial value, and is compared with the final value. What happens next depends
on whether to or downto is used:

1. In the case To is used, if the initial value is larger than the final value then Statement will
never be executed.
2. In the case DownTo is used, if the initial value is less than the final value then Statement
will never be executed.

After this check, the statement after Do is executed. After the execution of the statement, the control
variable is increased or decreased with 1, depending on whether To or Downto is used. The control
variable must be an ordinal type, no other types can be used as counters in a loop.
Remark: Contrary to ANSI pascal specifications, Free Pascal first initializes the counter variable, and only
then calculates the upper bound.
The following are valid loops:

For Day := Monday to Friday do Work;

For I := 100 downto 1 do
WriteLn (’Counting down : ’,i);
For I := 1 to 7*dwarfs do KissDwarf(i);

If the statement is a compound statement, then the Break (??) and Continue (??) reserved words
can be used to jump to the end or just after the end of the For statement.

9.2.5 The Repeat..until statement

The repeat statement is used to execute a statement until a certain condition is reached. The
statement will be executed at least once. The prototype syntax of the Repeat..until statement
is

Repeat statement

-
- repeat statement repeat statement until expression -
6 ;

82
 CHAPTER 9. STATEMENTS

This will execute the statements between repeat and until up to the moment when Expression
evaluates to True. Since the expression is evaluated after the execution of the statements, they
are executed at least once. Be aware of the fact that the boolean expression Expression will be
short-cut evaluated. (Meaning that the evaluation will be stopped at the point where the outcome is
known with certainty) The following are valid repeat statements

repeat
WriteLn (’I =’,i);
I := I+2;
until I>100;
repeat
X := X/2
until x<10e-3

The Break (??) and Continue (??) reserved words can be used to jump to the end or just after the
end of the repeat .. until statement.

9.2.6 The While..do statement

A while statement is used to execute a statement as long as a certain condition holds. This may
imply that the statement is never executed. The prototype syntax of the While..do statement is

While statements

-
- while statement while expression do statement -

This will execute Statement as long as Expression evaluates to True. Since Expression
is evaluated before the execution of Statement, it is possible that Statement isn’t executed at
all. Statement can be a compound statement. Be aware of the fact that the boolean expression
Expression will be short-cut evaluated. (Meaning that the evaluation will be stopped at the point
where the outcome is known with certainty) The following are valid while statements:

I := I+2;
while i<=100 do
begin
WriteLn (’I =’,i);
I := I+2;
end;
X := X/2;
while x>=10e-3 do
X := X/2;

They correspond to the example loops for the repeat statements.

If the statement is a compound statement, then the Break (??) and Continue (??) reserved words
can be used to jump to the end or just after the end of the While statement.

9.2.7 The With statement

The with statement serves to access the elements of a record or object or class, without having to
specify the name of the each time. The syntax for a with statement is

83
 CHAPTER 9. STATEMENTS

With statement

-
- with statement variable reference do statement -
6 ,

The variable reference must be a variable of a record, object or class type. In the with statement,
any variable reference, or method reference is checked to see if it is a field or method of the record
or object or class. If so, then that field is accessed, or that method is called. Given the declaration:

Type Passenger = Record

Name : String[30];
Flight : String[10];
end;
Var TheCustomer : Passenger;

The following statements are completely equivalent:

[Link] := ’Michael’;
[Link] := ’PS901’;

and

With TheCustomer do
begin
Name := ’Michael’;
Flight := ’PS901’;
end;

The statement

With A,B,C,D do Statement;

is equivalent to

With A do
With B do
With C do
With D do Statement;

This also is a clear example of the fact that the variables are tried last to first, i.e., when the compiler
encounters a variable reference, it will first check if it is a field or method of the last variable. If not,
then it will check the last-but-one, and so on. The following example shows this;

Program testw;
Type AR = record
X,Y : Longint;
end;
PAR = Record;

Var S,T : Ar;

begin
S.X := 1;S.Y := 1;

84
 CHAPTER 9. STATEMENTS

T.X := 2;T.Y := 2;
With S,T do
WriteLn (X,’ ’,Y);
end.

The output of this program is

2 2

Showing thus that the X,Y in the WriteLn statement match the T record variable.
Remark: When using a With statement with a pointer, or a class, it is not permitted to change the pointer or
the class in the With block. With the definitions of the previous example, the following illustrates
what it is about:

Var p : PAR;

begin
With P^ do
begin
// Do some operations
P:=OtherP;
X:=0.0; // Wrong X will be used !!
end;

The reason the pointer cannot be changed is that the address is stored by the compiler in a temporary
register. Changing the pointer won’t change the temporary address. The same is true for classes.

9.2.8 Exception Statements

Free Pascal supports exceptions. Exceptions provide a convenient way to program error and error-
recovery mechanisms, and are closely related to classes. Exception support is explained in chapter
13, page 113

9.3 Assembler statements

An assembler statement allows to insert assembler code right in the pascal code.

Assembler statements

-
- asm statement asm assembler code end -
registerlist

-
- registerlist [ stringconstant ] -
6 ,

More information about assembler blocks can be found in the Programmers guide. The register list is
used to indicate the registers that are modified by an assembler statement in the assembler block. The
compiler stores certain results in the registers. If the registers are modified in an assembler statement,

85
 CHAPTER 9. STATEMENTS

the compiler should, sometimes, be told about it. The registers are denoted with their Intel names
for the I386 processor, i.e., ’EAX’, ’ESI’ etc... As an example, consider the following assembler
code:

asm
Movl $1,%ebx
Movl $0,%eax
addl %eax,%ebx
end; [’EAX’,’EBX’];

This will tell the compiler that it should save and restore the contents of the EAX and EBX registers
when it encounters this asm statement.
Free Pascal supports various styles of assembler syntax. By default, AT&T syntax is assumed for the
80386 and compatibles platform. The default assembler style can be changed with the {$asmmode
xxx} switch in the code, or the -R command-line option. More about this can be found in the
Programmers guide.

86
 Chapter 10

Using functions and procedures

Free Pascal supports the use of functions and procedures, but with some extras: Function overloading
is supported, as well as Const parameters and open arrays.
Remark: In many of the subsequent paragraphs the words procedure and function will be used inter-
changeably. The statements made are valid for both, except when indicated otherwise.

10.1 Procedure declaration

A procedure declaration defines an identifier and associates it with a block of code. The procedure
can then be called with a procedure statement.

Procedure declaration

-
- procedure declaration procedure header ; subroutine block ; -

-
- procedure header procedure identifier -
qualified method identifier
- formal parameter list -
modifiers

-
- subroutine block block -
external directive
asm block
forward

See section 10.3, page 88 for the list of parameters. A procedure declaration that is followed by a
block implements the action of the procedure in that block. The following is a valid procedure :

Procedure DoSomething (Para : String);

begin
Writeln (’Got parameter : ’,Para);
Writeln (’Parameter in upper case : ’,Upper(Para));
end;

Note that it is possible that a procedure calls itself.

87
 CHAPTER 10. USING FUNCTIONS AND PROCEDURES

10.2 Function declaration

A function declaration defines an identifier and associates it with a block of code. The block of
code will return a result. The function can then be called inside an expression, or with a procedure
statement, if extended syntax is on.

Function declaration

-
- function declaration function header ; subroutine block ; -

-
- function header function identifier -
qualified method identifier
- formal parameter list : result type -
modifiers

-
- subroutine block block -
external directive
asm block
forward

The result type of a function can be any previously declared type. contrary to Turbo pascal, where
only simple types could be returned.

10.3 Parameter lists

When arguments must be passed to a function or procedure, these parameters must be declared in the
formal parameter list of that function or procedure. The parameter list is a declaration of identifiers
that can be referred to only in that procedure or function’s block.

Parameters

-
- formal parameter list ( parameter declaration ) -
6 ;

-
- parameter declaration value parameter -
constant parameter
variable parameter

Constant parameters and variable parameters can also be untyped parameters if they have no type
identifier.
As of version 1.1, Free Pascal supports default values for both constant parameters and value pa-
rameters, but only for simple types. The compiler must be in OBJFPC or DELPHI mode to accept
default values.

88
 CHAPTER 10. USING FUNCTIONS AND PROCEDURES

10.3.1 Value parameters

Value parameters are declared as follows:

Value parameters

-
- value parameter identifier list : parameter type -
array of
identifier : parameter type = default parameter value
- -

When parameters are declared as value parameters, the procedure gets a copy of the parameters that
the calling block passes. Any modifications to these parameters are purely local to the procedure’s
block, and do not propagate back to the calling block. A block that wishes to call a procedure with
value parameters must pass assignment compatible parameters to the procedure. This means that the
types should not match exactly, but can be converted (conversion code is inserted by the compiler
itself)
Care must be taken when using value parameters: Value parameters makes heavy use of the stack,
especially when using large parameters. The total size of all parameters in the formal parameter list
should be below 32K for portability’s sake (the Intel version limits this to 64K).
Open arrays can be passed as value parameters. See section 10.3.5, page 91 for more information on
using open arrays.
For a parameter of a simple type (i.e. not a structured type), a default value can be specified. This can
be an untyped constant. If the function call omits the parameter, the default value will be passed on
to the function. For dynamic arrays or other types that can be considered as equivalent to a pointer,
the only possible default value is Nil.
The following example will print 20 on the screen:

program testp;

Const
MyConst = 20;

Procedure MyRealFunc(I : Integer = MyConst);

begin
Writeln(’Function received : ’,I);
end;

begin
MyRealFunc;
end.

10.3.2 Variable parameters

Variable parameters are declared as follows:

Variable parameters

89
 CHAPTER 10. USING FUNCTIONS AND PROCEDURES

- variable parameter var identifier list

- -
: parameter type
array of

When parameters are declared as variable parameters, the procedure or function accesses immediatly
the variable that the calling block passed in its parameter list. The procedure gets a pointer to the
variable that was passed, and uses this pointer to access the variable’s value. From this, it follows that
any changes made to the parameter, will propagate back to the calling block. This mechanism can be
used to pass values back in procedures. Because of this, the calling block must pass a parameter of
exactly the same type as the declared parameter’s type. If it does not, the compiler will generate an
error.
Variable and constant parameters can be untyped. In that case the variable has no type, and hence
is incompatible with all other types. However, the address operator can be used on it, or it can be
can passed to a function that has also an untyped parameter. If an untyped parameter is used in an
assigment, or a value must be assigned to it, a typecast must be used.
File type variables must always be passed as variable parameters.
Open arrays can be passed as variable parameters. See section 10.3.5, page 91 for more information
on using open arrays.
Note that default values are not supported for variable parameters. This would make little sense since
it defeats the purpose of being able to pass a value back to the caller.

10.3.3 Out parameters

Out parameters (output parameters) are declared as follows:

Out parameters

-
- out parameter out identifier list -
: parameter type
array of

The purpose of an out parameter is to pass values back to the calling routine: The variable is passed
by reference. The initial value of the parameter on function entry is discarded, and should not be
used.
If a variable must be used to pass a value to a function and retrieve data from the function, then a
variable parameter must be used. If only a value must be retrieved, a out parameter can be used.
Needless to say, default values are not supported for out parameters.

10.3.4 Constant parameters

In addition to variable parameters and value parameters Free Pascal also supports Constant parame-
ters. A constant parameter as can be specified as follows:

Constant parameters

90
 CHAPTER 10. USING FUNCTIONS AND PROCEDURES

- constant parameter const

- identifier list -
: parameter type
array of
identifier : parameter type = default parameter value
- -

A constant argument is passed by reference if it’s size is larger than a pointer. It is passed by value if
the size is equal or is less then the size of a native pointer. This means that the function or procedure
receives a pointer to the passed argument, but it cannot be assigned to, this will result in a compiler
error. Furthermore a const parameter cannot be passed on to another function that requires a variable
parameter. The main use for this is reducing the stack size, hence improving performance, and still
retaining the semantics of passing by value...
Constant parameters can also be untyped. See section 10.3.2, page 89 for more information about
untyped parameters.
As for value parameters, constant parameters can get default values.
Open arrays can be passed as constant parameters. See section 10.3.5, page 91 for more information
on using open arrays.

10.3.5 Open array parameters

Free Pascal supports the passing of open arrays, i.e. a procedure can be declared with an array
of unspecified length as a parameter, as in Delphi. Open array parameters can be accessed in the
procedure or function as an array that is declared with starting index 0, and last element index
High(paremeter). For example, the parameter

Row : Array of Integer;

would be equivalent to

Row : Array[0..N-1] of Integer;

Where N would be the actual size of the array that is passed to the function. N-1 can be calculated
as High(Row). Open parameters can be passed by value, by reference or as a constant parameter.
In the latter cases the procedure receives a pointer to the actual array. In the former case, it receives a
copy of the array. In a function or procedure, open arrays can only be passed to functions which are
also declared with open arrays as parameters, not to functions or procedures which accept arrays of
fixed length. The following is an example of a function using an open array:

Function Average (Row : Array of integer) : Real;

Var I : longint;
Temp : Real;
begin
Temp := Row[0];
For I := 1 to High(Row) do
Temp := Temp + Row[i];
Average := Temp / (High(Row)+1);
end;

10.3.6 Array of const

In Object Pascal or Delphi mode, Free Pascal supports the Array of Const construction to pass
parameters to a subroutine.

91
 CHAPTER 10. USING FUNCTIONS AND PROCEDURES

This is a special case of the Open array construction, where it is allowed to pass any expression
in an array to a function or procedure.
In the procedure, passed the arguments can be examined using a special record:

Type
PVarRec = ^TVarRec;
TVarRec = record
case VType : Longint of
vtInteger : (VInteger: Longint);
vtBoolean : (VBoolean: Boolean);
vtChar : (VChar: Char);
vtExtended : (VExtended: PExtended);
vtString : (VString: PShortString);
vtPointer : (VPointer: Pointer);
vtPChar : (VPChar: PChar);
vtObject : (VObject: TObject);
vtClass : (VClass: TClass);
vtAnsiString : (VAnsiString: Pointer);
vtWideString : (VWideString: Pointer);
vtInt64 : (VInt64: PInt64);
end;

Inside the procedure body, the array of const is equivalent to an open array of TVarRec:

Procedure Testit (Args: Array of const);

Var I : longint;

begin
If High(Args)<0 then
begin
Writeln (’No aguments’);
exit;
end;
Writeln (’Got ’,High(Args)+1,’ arguments :’);
For i:=0 to High(Args) do
begin
write (’Argument ’,i,’ has type ’);
case Args[i].vtype of
vtinteger :
Writeln (’Integer, Value :’,args[i].vinteger);
vtboolean :
Writeln (’Boolean, Value :’,args[i].vboolean);
vtchar :
Writeln (’Char, value : ’,args[i].vchar);
vtextended :
Writeln (’Extended, value : ’,args[i].VExtended^);
vtString :
Writeln (’ShortString, value :’,args[i].VString^);
vtPointer :
Writeln (’Pointer, value : ’,Longint(Args[i].VPointer));
vtPChar :
Writeln (’PCHar, value : ’,Args[i].VPChar);
vtObject :

92
 CHAPTER 10. USING FUNCTIONS AND PROCEDURES

Writeln (’Object, name : ’,Args[i].[Link]);

vtClass :
Writeln (’Class reference, name :’,Args[i].[Link]);
vtAnsiString :
Writeln (’AnsiString, value :’,AnsiString(Args[I].VAnsiStr
else
Writeln (’(Unknown) : ’,args[i].vtype);
end;
end;
end;

In code, it is possible to pass an arbitrary array of elements to this procedure:

S:=’Ansistring 1’;
T:=’AnsiString 2’;
Testit ([]);
Testit ([1,2]);
Testit ([’A’,’B’]);
Testit ([TRUE,FALSE,TRUE]);
Testit ([’String’,’Another string’]);
Testit ([S,T]) ;
Testit ([P1,P2]);
Testit ([@testit,Nil]);
Testit ([ObjA,ObjB]);
Testit ([1.234,1.234]);
TestIt ([AClass]);

If the procedure is declared with the cdecl modifier, then the compiler will pass the array as a C
compiler would pass it. This, in effect, emulates the C construct of a variable number of arguments,
as the following example will show:

program testaocc;
{$mode objfpc}

Const
P : Pchar = ’example’;
Fmt : PChar =
’This %s uses printf to print numbers (%d) and strings.’#10;

// Declaration of standard C function printf:

procedure printf (fm : pchar; args : array of const);cdecl; external ’c’;

begin
printf(Fmt,[P,123]);
end.

Remark that this is not true for Delphi, so code relying on this feature will not be portable.

10.4 Function overloading

Function overloading simply means that the same function is defined more than once, but each time
with a different formal parameter list. The parameter lists must differ at least in one of it’s elements
type. When the compiler encounters a function call, it will look at the function parameters to decide

93
 CHAPTER 10. USING FUNCTIONS AND PROCEDURES

which one of the defined functions it should call. This can be useful when the same function must be
defined for different types. For example, in the RTL, the Dec procedure could be defined as:

...
Dec(Var I : Longint;decrement : Longint);
Dec(Var I : Longint);
Dec(Var I : Byte;decrement : Longint);
Dec(Var I : Byte);
...

When the compiler encounters a call to the dec function, it will first search which function it should
use. It therefore checks the parameters in a function call, and looks if there is a function definition
which matches the specified parameter list. If the compiler finds such a function, a call is inserted
to that function. If no such function is found, a compiler error is generated. functions that have a
cdecl modifier cannot be overloaded. (Technically, because this modifier prevents the mangling of
the function name by the compiler).
Prior to version 1.9 of the compiler, the overloaded functions needed to be in the same unit. Now the
compiler will continue searching in other units if it doesn’t find a matching version of an overloaded
function in one unit.
The compiler accepts the presence of the overload modifier as in Delphi, but it is not required,
unless in Delphi mode.

10.5 Forward defined functions

A function can be declared without having it followed by it’s implementation, by having it followed
by the forward procedure. The effective implementation of that function must follow later in
the module. The function can be used after a forward declaration as if it had been implemented
already. The following is an example of a forward declaration.

Program testforward;
Procedure First (n : longint); forward;
Procedure Second;
begin
WriteLn (’In second. Calling first...’);
First (1);
end;
Procedure First (n : longint);
begin
WriteLn (’First received : ’,n);
end;
begin
Second;
end.

A function can be defined as forward only once. Likewise, in units, it is not allowed to have a forward
declared function of a function that has been declared in the interface part. The interface declaration
counts as a forward declaration. The following unit will give an error when compiled:

Unit testforward;
interface
Procedure First (n : longint);
Procedure Second;

94
 CHAPTER 10. USING FUNCTIONS AND PROCEDURES

implementation
Procedure First (n : longint); forward;
Procedure Second;
begin
WriteLn (’In second. Calling first...’);
First (1);
end;
Procedure First (n : longint);
begin
WriteLn (’First received : ’,n);
end;
end.

10.6 External functions

The external modifier can be used to declare a function that resides in an external object file. It
allows to use the function in some code, and at linking time, the object file containing the implemen-
tation of the function or procedure must be linked in.

External directive

- external directive external

- -
string constant
name string constant
index integer constant

It replaces, in effect, the function or procedure code block. As an example:

program CmodDemo;
{$Linklib c}
Const P : PChar = ’This is fun !’;
Function strlen (P : PChar) : Longint; cdecl; external;
begin
WriteLn (’Length of (’,p,’) : ’,strlen(p))
end.

Remark: The parameters in our declaration of the external function should match exactly the ones in the
declaration in the object file.
If the external modifier is followed by a string constant:

external ’lname’;

Then this tells the compiler that the function resides in library ’lname’. The compiler will then
automatically link this library to the program.
The name that the function has in the library can also be specified:

external ’lname’ name ’Fname’;

This tells the compiler that the function resides in library ’lname’, but with name ’Fname’.The com-
piler will then automatically link this library to the program, and use the correct name for the function.
Under W INDOWS and OS /2, the following form can also be used:

95
 CHAPTER 10. USING FUNCTIONS AND PROCEDURES

external ’lname’ Index Ind;

This tells the compiler that the function resides in library ’lname’, but with index Ind. The compiler
will then automatically link this library to the program, and use the correct index for the function.
Finally, the external directive can be used to specify the external name of the function :

{$L myfunc.o}
external name ’Fname’;

This tells the compiler that the function has the name ’Fname’. The correct library or object file (in
this case myfunc.o) must still be linked. so that the function ’Fname’ is included in the linking stage.

10.7 Assembler functions

Functions and procedures can be completely implemented in assembly language. To indicate this,
use the assembler keyword:

Assembler functions

-
- asm block assembler ; declaration part asm statement -

Contrary to Delphi, the assembler keyword must be present to indicate an assembler function. For
more information about assembler functions, see the chapter on using assembler in the Programmers
guide.

10.8 Modifiers
A function or procedure declaration can contain modifiers. Here we list the various possibilities:

Modifiers

-
- modifiers ; public -
6 alias : string constant
interrupt
call modifiers

-
- call modifiers register -
pascal
cdecl
stdcall
popstack
saveregisters
inline
safecall
varargs

Free Pascal doesn’t support all Turbo Pascal modifiers, but does support a number of additional
modifiers. They are used mainly for assembler and reference to C object files.

96
 CHAPTER 10. USING FUNCTIONS AND PROCEDURES

10.8.1 alias
The alias modifier allows the programmer to specify a different name for a procedure or function.
This is mostly useful for referring to this procedure from assembly language constructs or from
another object file. As an example, consider the following program:

Program Aliases;

Procedure Printit;alias : ’DOIT’;

begin
WriteLn (’In Printit (alias : "DOIT")’);
end;
begin
asm
call DOIT
end;
end.

Remark: the specified alias is inserted straight into the assembly code, thus it is case sensitive.
The alias modifier does not make the symbol public to other modules, unless the routine is also
declared in the interface part of a unit, or the public modifier is used to force it as public. Consider
the following:

unit testalias;

interface

procedure testroutine;

implementation

procedure testroutine;alias:’ARoutine’;
begin
WriteLn(’Hello world’);
end;

end.

This will make the routine testroutine available publicly to external object files uunder the label
name ARoutine.

10.8.2 cdecl
The cdecl modifier can be used to declare a function that uses a C type calling convention. This
must be used when accessing functions residing in an object file generated by standard C compil-
ers. It allows to use the function in the code, and at linking time, the object file containing the C
implementation of the function or procedure must be linked in. As an example:

program CmodDemo;
{$LINKLIB c}
Const P : PChar = ’This is fun !’;
Function strlen (P : PChar) : Longint; cdecl; external name ’strlen’;
begin

97
 CHAPTER 10. USING FUNCTIONS AND PROCEDURES

WriteLn (’Length of (’,p,’) : ’,strlen(p))

end.

When compiling this, and linking to the C-library, the strlen function can be called throughout the
program. The external directive tells the compiler that the function resides in an external object
filebrary with the ’strlen’ name (see 10.6).
Remark: The parameters in our declaration of the C function should match exactly the ones in the declaration
in C.

10.8.3 export
The export modifier is used to export names when creating a shared library or an executable program.
This means that the symbol will be publicly available, and can be imported from other programs. For
more information on this modifier, consult the section on Programming dynamic libraries in the
Programmers guide.

10.8.4 inline
Procedures that are declared inline are copied to the places where they are called. This has the effect
that there is no actual procedure call, the code of the procedure is just copied to where the procedure
is needed, this results in faster execution speed if the function or procedure is used a lot.
By default, inline procedures are not allowed. Inline code must be enabled using the command-
line switch -Si or {$inline on} directive.

1. Inline code is NOT exported from a unit. This means that when calling an inline procedure
from another unit, a normal procedure call will be performed. Only inside units, Inline
procedures are really inlined.
2. Recursive inline functions are not allowed. i.e. an inline function that calls itself is not allowed.

10.8.5 interrupt
The interrupt keyword is used to declare a routine which will be used as an interrupt handler.
On entry to this routine, all the registers will be saved and on exit, all registers will be restored and
an interrupt or trap return will be executed (instead of the normal return from subroutine instruction).
On platforms where a return from interrupt does not exist, the normal exit code of routines will be
done instead. For more information on the generated code, consult the Programmers guide.

10.8.6 pascal
The pascal modifier can be used to declare a function that uses the classic pascal type calling
convention (passing parameters from left to right). For more information on the pascal calling con-
vention, consult the Programmers guide.

10.8.7 popstack
Popstack does the same as cdecl, namely it tells the Free Pascal compiler that a function uses the C
calling convention. In difference with the cdecl modifier, it still mangles the name of the function
as it would for a normal pascal function. With popstack, functions can be called by their pascal
names in a library.

98
 CHAPTER 10. USING FUNCTIONS AND PROCEDURES

10.8.8 public
The Public keyword is used to declare a function globally in a unit. This is useful if the function
should not be accessible from the unit file (i.e. another unit/program using the unit doesn’t see the
function), but must be accessible from the object file. as an example:

Unit someunit;
interface
Function First : Real;
Implementation
Function First : Real;
begin
First := 0;
end;
Function Second : Real; [Public];
begin
Second := 1;
end;
end.

If another program or unit uses this unit, it will not be able to use the function Second, since it
isn’t declared in the interface part. However, it will be possible to access the function Second at the
assembly-language level, by using it’s mangled name (see the Programmers guide).

10.8.9 register
The register keyword is used for compatibility with Delphi. In version 1.0.x of the compiler, this
directive has no effect on the generated code. As of the 1.9.X versions, this directive is supported.
The first three arguments are passed in registers EAX,ECX and EDX.

10.8.10 saveregisters
If this modifier is specified after a procedure or function, then the Free Pascal compiler will save all
registers on procedure entry, and restore them when the procedure exits (except for registers where
return values are stored).
This modifier is not used under normal circumstances, except maybe when calling assembler code.

10.8.11 safecall
This modifier ressembles closely the stdcall modifier. It sends parameters from right to left on
the stack. The called procedure saves and restores all registers.
More information about this modifier can be found in the Programmers guide, in the section on the
calling mechanism and the chapter on linking.

10.8.12 softfloat
This modifier makes sense only on the ARM architecture.

10.8.13 stdcall
This modifier pushes the parameters from right to left on the stack, it also aligns all the parameters
to a default alignment.

99
 CHAPTER 10. USING FUNCTIONS AND PROCEDURES

More information about this modifier can be found in the Programmers guide, in the section on the
calling mechanism and the chapter on linking.

10.8.14 varargs
This modifier can only be used together with the cdecl modifier, for external C procedures. It
indicates that the procedure accepts a variable number of arguments after the last declared variable.
These arguments are passed on without any type checking. It is equivalent to using the array of
const construction for cdecl procedures, without having to declare the array of const. The
square brackets around the variable arguments do not need to be used when this form of declaration
is used.

10.9 Unsupported Turbo Pascal modifiers

The modifiers that exist in Turbo pascal, but aren’t supported by Free Pascal, are listed in table (10.1).

Table 10.1: Unsupported modifiers

Modifier Why not supported ?

Near Free Pascal is a 32-bit compiler.
Far Free Pascal is a 32-bit compiler.

100
Chapter 11

Operator overloading

11.1 Introduction
Free Pascal supports operator overloading. This means that it is possible to define the action of some
operators on self-defined types, and thus allow the use of these types in mathematical expressions.
Defining the action of an operator is much like the definition of a function or procedure, only there
are some restrictions on the possible definitions, as will be shown in the subsequent.
Operator overloading is, in essence, a powerful notational tool; but it is also not more than that, since
the same results can be obtained with regular function calls. When using operator overloading, It is
important to keep in mind that some implicit rules may produce some unexpected results. This will
be indicated.

11.2 Operator declarations

To define the action of an operator is much like defining a function:

Operator definitions

-
- operator definition operator assignment operator definition -
arithmetic operator definition
comparision operator definition
- result identifier : result type ; subroutine block -

-
- assignment operator definition := ( value parameter ) -

-
- arithmetic operator definition + ( parameter list ) -
-
*
/
**

-
- comparision operator definition = ( parameter list ) -
<
<=
>
>=

101
 CHAPTER 11. OPERATOR OVERLOADING

The parameter list for a comparision operator or an arithmetic operator must always contain 2 pa-
rameters. The result type of the comparision operator must be Boolean.
Remark: When compiling in Delphi mode or Objfpc mode, the result identifier may be dropped. The
result can then be accessed through the standard Result symbol.
If the result identifier is dropped and the compiler is not in one of these modes, a syntax error will
occur.
The statement block contains the necessary statements to determine the result of the operation. It
can contain arbitrary large pieces of code; it is executed whenever the operation is encountered in
some expression. The result of the statement block must always be defined; error conditions are not
checked by the compiler, and the code must take care of all possible cases, throwing a run-time error
if some error condition is encountered.
In the following, the three types of operator definitions will be examined. As an example, throughout
this chapter the following type will be used to define overloaded operators on :

type
complex = record
re : real;
im : real;
end;

this type will be used in all examples.

The sources of the Run-Time Library contain a unit ucomplex, which contains a complete calculus
for complex numbers, based on operator overloading.

11.3 Assignment operators

The assignment operator defines the action of a assignent of one type of variable to another. The
result type must match the type of the variable at the left of the assignment statement, the single
parameter to the assignment operator must have the same type as the expression at the right of the
assignment operator.
This system can be used to declare a new type, and define an assignment for that type. For instance,
to be able to assign a newly defined type ’Complex’

Var
C,Z : Complex; // New type complex

begin
Z:=C; // assignments between complex types.
end;

The following assignment operator would have to be defined:

Operator := (C : Complex) z : complex;

To be able to assign a real type to a complex type as follows:

var
R : real;
C : complex;

102
 CHAPTER 11. OPERATOR OVERLOADING

begin
C:=R;
end;

the following assignment operator must be defined:

Operator := (r : real) z : complex;

As can be seen from this statement, it defines the action of the operator := with at the right a real
expression, and at the left a complex expression.
an example implementation of this could be as follows:

operator := (r : real) z : complex;

begin
[Link]:=r;
[Link]:=0.0;
end;

As can be seen in the example, the result identifier (z in this case) is used to store the result of
the assignment. When compiling in Delphi mode or objfpc mode, the use of the special identifier
Result is also allowed, and can be substituted for the z, so the above would be equivalent to

operator := (r : real) z : complex;

begin
[Link]:=r;
[Link]:=0.0;
end;

The assignment operator is also used to convert types from one type to another. The compiler will
consider all overloaded assignment operators till it finds one that matches the types of the left hand
and right hand expressions. If no such operator is found, a ’type mismatch’ error is given.
Remark: The assignment operator is not commutative; the compiler will never reverse the role of the two
arguments. in other words, given the above definition of the assignment operator, the following is
not possible:

var
R : real;
C : complex;

begin
R:=C;
end;

if the reverse assignment should be possible (this is not so for reals and complex numbers) then the
assigment operator must be defined for that as well.
Remark: The assignment operator is also used in implicit type conversions. This can have unwanted effects.
Consider the following definitions:

operator := (r : real) z : complex;

function exp(c : complex) : complex;

103
 CHAPTER 11. OPERATOR OVERLOADING

then the following assignment will give a type mismatch:

Var
r1,r2 : real;

begin
r1:=exp(r2);
end;

because the compiler will encounter the definition of the exp function with the complex argument. It
implicitly converts r2 to a complex, so it can use the above exp function. The result of this function
is a complex, which cannot be assigned to r1, so the compiler will give a ’type mismatch’ error. The
compiler will not look further for another exp which has the correct arguments.
It is possible to avoid this particular problem by specifying

r1:=[Link](r2);

An experimental solution for this problem exists in the compiler, but is not enabled by default. Maybe
someday it will be.

11.4 Arithmetic operators

Arithmetic operators define the action of a binary operator. Possible operations are:

multiplication to multiply two types, the * multiplication operator must be overloaded.

division to divide two types, the / division operator must be overloaded.
addition to add two types, the + addition operator must be overloaded.
substraction to substract two types, the - substraction operator must be overloaded.
exponentiation to exponentiate two types, the ** exponentiation operator must be overloaded.

The definition of an arithmetic operator takes two parameters. The first parameter must be of the
type that occurs at the left of the operator, the second parameter must be of the type that is at the
right of the arithmetic operator. The result type must match the type that results after the arithmetic
operation.
To compile an expression as

var
R : real;
C,Z : complex;

begin
C:=R*Z;
end;

one needs a definition of the multiplication operator as:

Operator * (r : real; z1 : complex) z : complex;

begin
[Link] := [Link] * r;
[Link] := [Link] * r;
end;

104
 CHAPTER 11. OPERATOR OVERLOADING

As can be seen, the first operator is a real, and the second is a complex. The result type is complex.
Multiplication and addition of reals and complexes are commutative operations. The compiler, how-
ever, has no notion of this fact so even if a multiplication between a real and a complex is defined,
the compiler will not use that definition when it encounters a complex and a real (in that order). It is
necessary to define both operations.
So, given the above definition of the multiplication, the compiler will not accept the following state-
ment:

var
R : real;
C,Z : complex;

begin
C:=Z*R;
end;

since the types of Z and R don’t match the types in the operator definition.
The reason for this behaviour is that it is possible that a multiplication is not always commutative.
e.g. the multiplication of a (n,m) with a (m,n) matrix will result in a (n,n) matrix, while the
mutiplication of a (m,n) with a (n,m) matrix is a (m,m) matrix, which needn’t be the same in all
cases.

11.5 Comparision operator

The comparision operator can be overloaded to compare two different types or to compare two equal
types that are not basic types. The result type of a comparision operator is always a boolean.
The comparision operators that can be overloaded are:

equal to (=) to determine if two variables are equal.

less than (<) to determine if one variable is less than another.
greater than (>) to determine if one variable is greater than another.
greater than or equal to (>=) to determine if one variable is greater than or equal to another.
less than or equal to (<=) to determine if one variable is greater than or equal to another.

There is no separate operator for unequal to (<>). To evaluate a statement that contans the unequal
to operator, the compiler uses the equal to operator (=), and negates the result.
As an example, the following opetrator allows to compare two complex numbers:

operator = (z1, z2 : complex) b : boolean;

the above definition allows comparisions of the following form:

Var
C1,C2 : Complex;

begin
If C1=C2 then
Writeln(’C1 and C2 are equal’);
end;

105
 CHAPTER 11. OPERATOR OVERLOADING

The comparision operator definition needs 2 parameters, with the types that the operator is meant to
compare. Here also, the compiler doesn’t apply commutativity; if the two types are different, then it
necessary to define 2 comparision operators.
In the case of complex numbers, it is, for instance necessary to define 2 comparsions: one with the
complex type first, and one with the real type first.
Given the definitions

operator = (z1 : complex;r : real) b : boolean;

operator = (r : real; z1 : complex) b : boolean;

the following two comparisions are possible:

Var
R,S : Real;
C : Complex;

begin
If (C=R) or (S=C) then
Writeln (’Ok’);
end;

Note that the order of the real and complex type in the two comparisions is reversed.

106
Chapter 12

Programs, units, blocks

A Pascal program consists of modules called units. A unit can be used to group pieces of code
together, or to give someone code without giving the sources. Both programs and units consist of
code blocks, which are mixtures of statements, procedures, and variable or type declarations.

12.1 Programs
A pascal program consists of the program header, followed possibly by a ’uses’ clause, and a block.

Programs

-
- program program header ; block . -
uses clause

-
- program header program identifier -
( program parameters )

-
- program parameters identifier list -

-
- uses clause uses identifier ; -
6 ,

The program header is provided for backwards compatibility, and is ignored by the compiler. The
uses clause serves to identify all units that are needed by the program. The system unit doesn’t have
to be in this list, since it is always loaded by the compiler. The order in which the units appear is
significant, it determines in which order they are initialized. Units are initialized in the same order as
they appear in the uses clause. Identifiers are searched in the opposite order, i.e. when the compiler
searches for an identifier, then it looks first in the last unit in the uses clause, then the last but one,
and so on. This is important in case two units declare different types with the same identifier. When
the compiler looks for unit files, it adds the extension .ppu (.ppw for Win32 platforms) to the name
of the unit. On LINUX and in operating systems where filenames are case sensitive, when looking
for a unit, the unit name is first looked for in the original case, and when not found, converted to all
lowercase and searched for.
If a unit name is longer than 8 characters, the compiler will first look for a unit name with this length,
and then it will truncate the name to 8 characters and look for it again. For compatibility reasons,
this is also true on platforms that suport long file names.

107
 CHAPTER 12. PROGRAMS, UNITS, BLOCKS

12.2 Units
A unit contains a set of declarations, procedures and functions that can be used by a program or
another unit. The syntax for a unit is as follows:

Units

-
- unit unit header interface part implementation part -
- end . -
initialization part
finalization part
begin statement
6 ;

-
- unit header unit unit identifier ; -

-
- interface part interface -
uses clause 6 constant declaration part
type declaration part
procedure headers part

-
- procedure headers part procedure header ; -
function header call modifiers ;

-
- implementation part implementation declaration part -
uses clause

-
- initialization part initialization statement -
6 ;

-
- finalization part finalization statement -
6 ;

The interface part declares all identifiers that must be exported from the unit. This can be constant,
type or variable identifiers, and also procedure or function identifier declarations. Declarations inside
the implementation part are not accessible outside the unit. The implementation must contain a
function declaration for each function or procedure that is declared in the interface part. If a function
is declared in the interface part, but no declaration of that function is present in the implementation
part, then the compiler will give an error.
When a program uses a unit (say unitA) and this units uses a second unit, say unitB, then the program
depends indirectly also on unitB. This means that the compiler must have access to unitB when
trying to compile the program. If the unit is not present at compile time, an error occurs.
Note that the identifiers from a unit on which a program depends indirectly, are not accessible to the
program. To have access to the identifiers of a unit, the unit must be in the uses clause of the program
or unit where the identifiers are needed.
Units can be mutually dependent, that is, they can reference each other in their uses clauses. This is
allowed, on the condition that at least one of the references is in the implementation section of the
unit. This also holds for indirect mutually dependent units.
If it is possible to start from one interface uses clause of a unit, and to return there via uses clauses
of interfaces only, then there is circular unit dependence, and the compiler will generate an error. As
and example : the following is not allowed:

108
 CHAPTER 12. PROGRAMS, UNITS, BLOCKS

Unit UnitA;
interface
Uses UnitB;
implementation
end.

Unit UnitB
interface
Uses UnitA;
implementation
end.

But this is allowed :

Unit UnitA;
interface
Uses UnitB;
implementation
end.
Unit UnitB
implementation
Uses UnitA;
end.

Because UnitB uses UnitA only in it’s implentation section. In general, it is a bad idea to have
circular unit dependencies, even if it is only in implementation sections.

12.3 Blocks
Units and programs are made of blocks. A block is made of declarations of labels, constants, types
variables and functions or procedures. Blocks can be nested in certain ways, i.e., a procedure or
function declaration can have blocks in themselves. A block looks like the following:

Blocks

-
- block declaration part statement part -

-
- declaration part -
6 label declaration part
constant declaration part
resourcestring declaration part
type declaration part
variable declaration part
threadvariable declaration part
procedure/function declaration part

-
- label declaration part label label ; -
6,

-
- constant declaration part const constant declaration -
6 typed constant declaration

109
 CHAPTER 12. PROGRAMS, UNITS, BLOCKS

-
- resourcestring declaration part resourcestring string constant declaration -
6
- -

-
- type declaration part type type declaration -
6

-
- variable declaration part var variable declaration -
6

-
- threadvariable declaration part threadvar variable declaration -
6

-
- procedure/function declaration part procedure declaration -
6 function declaration
constructor declaration
destructor declaration

-
- statement part compound statement -

Labels that can be used to identify statements in a block are declared in the label declaration part
of that block. Each label can only identify one statement. Constants that are to be used only in one
block should be declared in that block’s constant declaration part. Variables that are to be used only
in one block should be declared in that block’s constant declaration part. Types that are to be used
only in one block should be declared in that block’s constant declaration part. Lastly, functions and
procedures that will be used in that block can be declared in the procedure/function declaration part.
After the different declaration parts comes the statement part. This contains any actions that the block
should execute. All identifiers declared before the statement part can be used in that statement part.

12.4 Scope
Identifiers are valid from the point of their declaration until the end of the block in which the dec-
laration occurred. The range where the identifier is known is the scope of the identifier. The exact
scope of an identifier depends on the way it was defined.

12.4.1 Block scope

The scope of a variable declared in the declaration part of a block, is valid from the point of declara-
tion until the end of the block. If a block contains a second block, in which the identfier is redeclared,
then inside this block, the second declaration will be valid. Upon leaving the inner block, the first
declaration is valid again. Consider the following example:

Program Demo;
Var X : Real;
{ X is real variable }
Procedure NewDeclaration
Var X : Integer; { Redeclare X as integer}
begin
// X := 1.234; {would give an error when trying to compile}
X := 10; { Correct assigment}

110
 CHAPTER 12. PROGRAMS, UNITS, BLOCKS

end;
{ From here on, X is Real again}
begin
X := 2.468;
end.

In this example, inside the procedure, X denotes an integer variable. It has it’s own storage space,
independent of the variable X outside the procedure.

12.4.2 Record scope

The field identifiers inside a record definition are valid in the following places:

1. to the end of the record definition.

2. field designators of a variable of the given record type.
3. identifiers inside a With statement that operates on a variable of the given record type.

12.4.3 Class scope

A component identifier is valid in the following places:

1. From the point of declaration to the end of the class definition.

2. In all descendent types of this class, unless it is in the private part of the class declaration.
3. In all method declaration blocks of this class and descendent classes.
4. In a with statement that operators on a variable of the given class’s definition.

Note that method designators are also considered identifiers.

12.4.4 Unit scope

All identifiers in the interface part of a unit are valid from the point of declaration, until the end
of the unit. Furthermore, the identifiers are known in programs or units that have the unit in their
uses clause. Identifiers from indirectly dependent units are not available. Identifiers declared in the
implementation part of a unit are valid from the point of declaration to the end of the unit. The system
unit is automatically used in all units and programs. It’s identifiers are therefore always known, in
each pascal program, library or unit. The rules of unit scope imply that an identifier of a unit can
be redefined. To have access to an identifier of another unit that was redeclared in the current unit,
precede it with that other units name, as in the following example:

unit unitA;
interface
Type
MyType = Real;
implementation
end.
Program prog;
Uses UnitA;

{ Redeclaration of MyType}

111
 CHAPTER 12. PROGRAMS, UNITS, BLOCKS

Type MyType = Integer;

Var A : Mytype; { Will be Integer }
B : [Link] { Will be real }
begin
end.

This is especially useful when redeclaring the system unit’s identifiers.

12.5 Libraries
Free Pascal supports making of dynamic libraries (DLLs under Win32 and OS /2) trough the use of
the Library keyword.
A Library is just like a unit or a program:

Libraries

-
- library library header ; block . -
uses clause

-
- library header library identifier -

By default, functions and procedures that are declared and implemented in library are not available
to a programmer that wishes to use this library.
In order to make functions or procedures available from the library, they must be exported in an
export clause:

Exports clause

-
- exports clause exports exports list ; -

-
- exports list exports entry -
6 ,

- exports entry identifier

- -
index integer constant name string constant
- -

Under Win32, an index clause can be added to an exports entry. an index entry must be a positive
number larger or equal than 1.
Optionally, an exports entry can have a name specifier. If present, the name specifier gives the exact
name (case sensitive) of the function in the library.
If neither of these constructs is present, the functions or procedures are exported with the exact names
as specified in the exports clause.

112
 Chapter 13

Exceptions

Exceptions provide a convenient way to program error and error-recovery mechanisms, and are
closely related to classes. Exception support is based on 3 constructs:

Raise statements. To raise an exeption. This is usually done to signal an error condition.

Try ... Except blocks. These block serve to catch exceptions raised within the scope of the block,
and to provide exception-recovery code.
Try ... Finally blocks. These block serve to force code to be executed irrespective of an exception
occurrence or not. They generally serve to clean up memory or close files in case an exception
occurs. The compiler generates many implicit Try ... Finally blocks around proce-
dure, to force memory consistence.

13.1 The raise statement

The raise statement is as follows:

Raise statement

-
- raise statement -
exception instance
at address expression

This statement will raise an exception. If it is specified, the exception instance must be an initialized
instance of a class, which is the raise type. The address exception is optional. If itis not specified,
the compiler will provide the address by itself. If the exception instance is omitted, then the current
exception is re-raised. This construct can only be used in an exception handling block (see further).
Remark: Control never returns after an exception block. The control is transferred to the first try...finally
or try...except statement that is encountered when unwinding the stack. If no such statement
is found, the Free Pascal Run-Time Library will generate a run-time error 217 (see also section 13.5,
page 116).
As an example: The following division checks whether the denominator is zero, and if so, raises an
exception of type EDivException

113
 CHAPTER 13. EXCEPTIONS

Type EDivException = Class(Exception);

Function DoDiv (X,Y : Longint) : Integer;
begin
If Y=0 then
Raise [Link] (’Division by Zero would occur’);
Result := X Div Y;
end;

The class Exception is defined in the Sysutils unit of the rtl. (section 13.5, page 116)

13.2 The try...except statement

A try...except exception handling block is of the following form :

Try..except statement

-
- try statement try statement list except exceptionhandlers end -

-
- statement list statement -
6 ;

-
- exceptionhandlers -
exception handler
6 ; else statement list
statement list

- exception handler on
- class type identifier do statement -
identifier :

If no exception is raised during the execution of the statement list, then all statements in the
list will be executed sequentially, and the except block will be skipped, transferring program flow to
the statement after the final end.
If an exception occurs during the execution of the statement list, the program flow will be
transferred to the except block. Statements in the statement list between the place where the exception
was raised and the exception block are ignored.
In the exception handling block, the type of the exception is checked, and if there is an exception
handler where the class type matches the exception object type, or is a parent type of the exception
object type, then the statement following the corresponding Do will be executed. The first matching
type is used. After the Do block was executed, the program continues after the End statement.
The identifier in an exception handling statement is optional, and declares an exception object. It
can be used to manipulate the exception object in the exception handling code. The scope of this
declaration is the statement block foillowing the Do keyword.
If none of the On handlers matches the exception object type, then the statement list after else is
executed. If no such list is found, then the exception is automatically re-raised. This process allows
to nest try...except blocks.
If, on the other hand, the exception was caught, then the exception object is destroyed at the end of
the exception handling block, before program flow continues. The exception is destroyed through a
call to the object’s Destroy destructor.
As an example, given the previous declaration of the DoDiv function, consider the following

114
 CHAPTER 13. EXCEPTIONS

Try
Z := DoDiv (X,Y);
Except
On EDivException do Z := 0;
end;

If Y happens to be zero, then the DoDiv function code will raise an exception. When this happens,
program flow is transferred to the except statement, where the Exception handler will set the value
of Z to zero. If no exception is raised, then program flow continues past the last end statement. To
allow error recovery, the Try ... Finally block is supported. A Try...Finally block
ensures that the statements following the Finally keyword are guaranteed to be executed, even if
an exception occurs.

13.3 The try...finally statement

A Try..Finally statement has the following form:

Try...finally statement

-
- trystatement try statement list finally finally statements end -

-
- finally statements statementlist -

If no exception occurs inside the statement List, then the program runs as if the Try, Finally
and End keywords were not present.
If, however, an exception occurs, the program flow is immediatly transferred from the point where
the excepion was raised to the first statement of the Finally statements.
All statements after the finally keyword will be executed, and then the exception will be automatically
re-raised. Any statements between the place where the exception was raised and the first statement
of the Finally Statements are skipped.
As an example consider the following routine:

Procedure Doit (Name : string);

Var F : Text;
begin
Try
Assign (F,Name);
Rewrite (name);
... File handling ...
Finally
Close(F);
end;

If during the execution of the file handling an execption occurs, then program flow will continue at
the close(F) statement, skipping any file operations that might follow between the place where
the exception was raised, and the Close statement. If no exception occurred, all file operations will
be executed, and the file will be closed at the end.

115
 CHAPTER 13. EXCEPTIONS

13.4 Exception handling nesting

It is possible to nest Try...Except blocks with Try...Finally blocks. Program flow will be
done according to a lifo (last in, first out) principle: The code of the last encountered Try...Except
or Try...Finally block will be executed first. If the exception is not caught, or it was a finally
statement, program flow will be transferred to the last-but-one block, ad infinitum.
If an exception occurs, and there is no exception handler present, then a runerror 217 will be gen-
erated. When using the sysutils unit, a default handler is installed which will show the exception
object message, and the address where the exception occurred, after which the program will exit with
a Halt instruction.

13.5 Exception classes

The sysutils unit contains a great deal of exception handling. It defines the following exception
types:

Exception = class(TObject)
private
fmessage : string;
fhelpcontext : longint;
public
constructor create(const msg : string);
constructor createres(indent : longint);
property helpcontext : longint read fhelpcontext write fhelpcontext;
property message : string read fmessage write fmessage;
end;
ExceptClass = Class of Exception;
{ mathematical exceptions }
EIntError = class(Exception);
EDivByZero = class(EIntError);
ERangeError = class(EIntError);
EIntOverflow = class(EIntError);
EMathError = class(Exception);

The sysutils unit also installs an exception handler. If an exception is unhandled by any exception
handling block, this handler is called by the Run-Time library. Basically, it prints the exception
address, and it prints the message of the Exception object, and exits with a exit code of 217. If the
exception object is not a descendent object of the Exception object, then the class name is printed
instead of the exception message.
It is recommended to use the Exception object or a descendant class for all raise statements,
since then the message field of the exception object can be used.

116
Chapter 14

Using assembler

Free Pascal supports the use of assembler in code, but not inline assembler macros. To have more
information on the processor specific assembler syntax and its limitations, see the Programmers
guide.

14.1 Assembler statements

The following is an example of assembler inclusion in pascal code.

...
Statements;
...
Asm
the asm code here
...
end;
...
Statements;

The assembler instructions between the Asm and end keywords will be inserted in the assembler
generated by the compiler. Conditionals can be used ib assembler, the compiler will recognise it, and
treat it as any other conditionals.

14.2 Assembler procedures and functions

Assembler procedures and functions are declared using the Assembler directive. This permits the
code generator to make a number of code generation optimizations.
The code generator does not generate any stack frame (entry and exit code for the routine) if it
contains no local variables and no parameters. In the case of functions, ordinal values must be
returned in the accumulator. In the case of floating point values, these depend on the target processor
and emulation options.

117