ABL Reference

Data types

The data type of a data element defines what kind of data the data element can store. ABL supports the following basic kinds of data types:

ABL built-in primitive types, including mappings to corresponding .NET primitive types

Object types, which include both ABL and supported .NET object types, including both built-in and user-defined class, enumeration, and interface types

ABL handle-based objects

ABL arrays, including one-dimensional arrays of ABL primitive types, ABL object types, or .NET object types

ABL primitive types (see Table 21) are built-in data types that can hold values with relatively simple content and that support relatively simple operations that can typically be applied using built-in ABL operands, such as arithmetic or relational operands (for example, the + Addition operator, or the EQ or = operator). The values for all ABL primitive types, other than the MEMPTR, BLOB, CLOB, and LONGCHAR, are limited in memory size to 32KB. MEMPTR and LONGCHAR variables can be any size, and BLOB and CLOB fields can be up to 1GB in size. Note that you can define BLOB and CLOB data types only for table or temp-table fields, and, unlike most other ABL primitive types, the operations they support do not have built-in ABL operands, but are available using built-in ABL functions (for example, the COPY-LOB statement).

You can use a primitive data type keyword in the following ABL syntax:

DEFINE PARAMETER statement

DEFINE PROPERTY statement

DEFINE VARIABLE statement

Fields of a table from an OpenEdge RDBMS using Progress Developer Studio for OpenEdge or the Data Dictionary.

Fields of a temp-table using the DEFINE TEMP-TABLE statement or methods of a Temp-table object handle.

FUNCTION statement (return type)

METHOD statement (return type)

Parameter definition syntax (for a user-defined method or function)

ABL object types (see Table 22) are complex types that function according to object-oriented principles. They can be broadly divided into two categories:

Classes — ABL supports a set of built-in object types and also allows you to create your own user-defined object types using using the CLASS statement and the INTERFACE statement. All ABL class types ultimately inherit from the ABL root class, Progress.Lang.Object class. Each object type encapsulates a set of data and behavioral elements (members). ABL class members can include implementations for data members, properties, and methods.

You can create instances of classes (class-based objects) at run time using the NEW function (classes), and you can reference each instance and its PUBLIC members using an object reference, which is an ABL data element defined to reference a specific object type. You can define an object reference for any kind of ABL data element that you can define as an ABL primitive type (except a database table field). However, you can define a field of a temp-table as an object reference to the ABL root class (Progress.Lang.Object class).

Enumerations — An enumeration (also known as enumerated types or enums) is made up of a list of strongly typed, named constants called members. The value of a variable defined as an enumeration is restricted to the list of members defined for that enumeration. As with classes, ABL supports a set of built-in enumeration types and also allows you to create your own user-defined enumerations using the ENUM statement. You can define two types of enum. In the standard version, a variable defined as an enum can only be assigned one value at a time. You can also define a flag enum. Any variable defined as a flag enum can be assigned any combination of the enum's members at one time.

Enumerations are implicitly class based, but unlike classes defined using the CLASS statement, enumeration types cannot be explicitly instantiated using the NEW function. To assign a value to a variable, property, or parameter defined as an enumeration type, you use Type-name syntax to access the members defined in the enumeration type. See the Enumeration member access reference entry for more information.

For information on the built-in ABL class, interface, and enumeration types, see the Class, Interface, and Enumeration Reference.

.NET primitive types (see Table 24) include data types that are built-into .NET languages. (Note that each language uses its own nomenclature. Compare, for example, the C# int versus the Visual Basic Integer.) .NET also aliases (maps) a standard set of object types (.NET mapped object types, for example, System.Int32) to these primitive types. All .NET languages can reference each such type as either the primitive type that the particular language supports or as the alias for the corresponding .NET object type that every language supports.

ABL also references both the .NET primitive types and their corresponding mapped object types by mapping each ABL primitive type to a given set of .NET primitive and object type mappings (see Table 24). Thus, ABL documentation refers collectively to both the .NET primitive types and the corresponding .NET mapped object types as .NET mapped data types.

.NET object types (see Table 22) include all class types (and their derivatives) that derive from the .NET root class, System.Object, and .NET interface types, which can inherit from other .NET interface types, but otherwise function for a class much like an ABL interface type. You can reference a .NET object type like an ABL object type, by using an ABL object reference defined as that object type, or by referencing members of a .NET class. .NET object types also consist of two basic kinds of types:

Value types — Objects that .NET creates, passes, and assigns by value. Value type objects all inherit from the .NET class, System.ValueType. In ABL, when you access a value type from .NET, you access a new copy of the object that is separate from the one that is maintained by .NET. If you then change object data in ABL, these changes do not appear in any copy of the object maintained by .NET. In addition, the ABL object reference to the ABL copy of the object has no affect on the .NET garbage collection of any .NET copy of the object.

Reference types — Objects that .NET creates, passes, and assigns by reference. In ABL, when you access a .NET reference type, you access the same copy of the object that is maintained by .NET. If you then change object data in ABL, these changes also appear in .NET, because .NET references the same object. In addition, the ABL object reference is counted as a reference to the object for .NET garbage collection.

ABL also provides limited support for .NET abstract classes and .NET generic types. A .NET abstract class is similar to an ABL abstract class. A generic type has a type definition that can function as one of several different types, depending on type parameters used to complete the effective type name. .NET generic types are briefly described further in this entry. For more information on the basic kinds of .NET object types that ABL supports, see the notes section of this reference entry. OpenEdge also provides a set of built-in .NET class and interface types to support access to .NET object types. For information on the built-in .NET class and interface types, see the Class, Interface, and Enumeration Reference.

Within certain restrictions, an ABL class can inherit from a .NET class and implement .NET interfaces, similar to inheriting from an ABL class or implementing an ABL interface, respectively. When an ABL class inherits from a .NET class, any of its methods that override methods in the .NET class hierarchy can be called polymorphically on the .NET super class from both ABL and .NET. As a result, any ABL class that inherits from a .NET class becomes an ABL-derived .NET class. In fact, when an ABL-derived .NET class is instantiated in an ABL session, an instance with a corresponding .NET class type is also instantiated in the .NET context with reference to any ABL-overridden methods.

Similarly, when an ABL class implements a .NET interface, all of the ABL-implemented properties and methods can be called polymorphically on the interface type from both ABL and .NET. As a result, any ABL class that implements a .NET interface becomes an ABL-extended .NET class. In fact, when an ABL-extended .NET class that implements .NET interfaces is instantiated in an ABL session, an instance with a corresponding .NET class type is also instantiated in the .NET context with reference to the ABL-implemented properties and methods that might be accessed from .NET on each of the implemented interface types.

Note: An ABL-derived .NET class is also considered an ABL-extended .NET class. However, an ABL-extended .NET class that only implements .NET interfaces is not an ABL-derived .NET class. Thus, when OpenEdge documentation refers to an ABL-extended .NET class, it can also (but not necessarily) be referring to an ABL-derived .NET class.

ABL handle-based objects (see Table 22) include a set of complex, weakly-typed objects, some of which ABL provides as built-in system objects (such as, the SESSION system handle), and others that ABL supports as a pre-defined set of objects that you can create as needed (such as, a FRAME widget or a record buffer). Handle-based objects exist independently and have no inheritance hierarchy like class-based objects. However, like class-based objects, handle-based objects have members consisting of a set of attributes (data) and methods (behavior). Depending on the object, you can create a handle-based object as a compile-time (static) object using an appropriate DEFINE statement or as a run-time (dynamic) object using an appropriate CREATE statement or other executable statement. You can reference system objects using the built-in system handle (keyword) pre-defined for them. You can reference static handle-based objects by name, using appropriate syntax for each type, and you can reference dynamic or static handle-based objects using a common primitive data element known as a handle, which you define as the HANDLE data type. Because of the weak typing of these objects, you can reference all static and dynamic handle-based objects that you define or create using the same handle. ABL also provides some system handles that provide access to particular types of pre-defined handle-based objects that are in a given state (such as, the CURRENT-WINDOW system handle for accessing a particular WINDOW widget).

ABL arrays (see Table 22) are limited to one dimension and can include elements of any primitive or object type that you can define for a variable (see the DEFINE VARIABLE statement reference entry). ABL also provides support for mapping ABL arrays to one-dimensional .NET array objects of the same element type. This means that while you can access all .NET arrays as class instances, you can also make direct array assignments and pass routine parameters between ABL arrays and equivalent one-dimensional .NET array objects. This also includes .NET arrays whose elements are .NET mapped data types (referred to as .NET arrays of mapped types), where assignments to or from ABL arrays work using the rules of implicit data type mapping. The element types supported for a .NET array of mapped types are identical to the .NET data types that ABL implicitly maps to ABL primitive types (see Table 24). For more information on support for both ABL arrays and .NET arrays, see the notes section of this reference entry.

The following table describes the primitive types supported in ABL.

Table 21. ABL primitive types
Primitive type	Description
BLOB	BLOB (Binary Large OBject) specifies a database table or temp-table field that contains a BLOB locator, which points to the associated BLOB data stored in the database. You must use a MEMPTR to manipulate the binary contents of a BLOB field in ABL.
CHARACTER	CHARACTER data consists of numbers, letters, and special characters.
CLOB	CLOB (Character Large OBject) specifies a database table or temp-table field that contains a CLOB locator, which points to the associated CLOB data stored in the database. You must use a LONGCHAR to manipulate the character contents of a CLOB field in ABL.
COM-HANDLE	A COM-HANDLE is a handle to a COM object (ActiveX Automation object or ActiveX Control).
DATE	DATE fields contain dates.
DATETIME	DATETIME data has two parts: an ABL date and an ABL time. The unit of time is milliseconds from midnight.
DATETIME-TZ	DATETIME-TZ data has three parts: an ABL date, an ABL time, and an integer representing the time zone offset from Coordinated Universal Time (UTC). The unit of time is milliseconds from midnight. The unit of time zone offset is minutes.
DECIMAL	DECIMAL data consists of decimal numbers up to 50 digits in length including up to 10 digits to the right of the decimal point.
HANDLE	A HANDLE is a pointer to an ABL handle-based object. This can be a compile-time defined static object or a run-time defined dynamic object. Note: HANDLE and WIDGET-HANDLE can be assigned to each other and used interchangeably. WIDGET-HANDLE is supported only for backward compatibility.
INT64	An INT64 consists of 64-bit data (whole numbers).
INTEGER	An INTEGER consists of 32-bit data (whole numbers).
LOGICAL	LOGICAL data evaluates to TRUE or FALSE (or YES or NO).
LONGCHAR	A LONGCHAR consists of CHARACTER data that is not limited to 32K in size. You can use a LONGCHAR to manipulate the character contents of a CLOB database or temp-table field in ABL.
MEMPTR	A MEMPTR contains a sequence of bytes in memory. You can use a MEMPTR to manipulate the contents of a BLOB database or temp-table field in ABL.
RAW	RAW data can be any kind of data, even data from non-OpenEdge databases. It is not converted in any way.
RECID	A RECID is a unique internal identifier for a record within a single database storage area. Note: RECID is supported mainly for backward compatibility. For most applications, use ROWID instead.
ROWID	A ROWID is a unique internal identifier for a record within a single database storage area. For partitioned tables, it is also unique across all partitions of a given table.
WIDGET-HANDLE	A WIDGET-HANDLE is a pointer to an ABL handle-based object. Note: HANDLE and WIDGET-HANDLE can be assigned to each other and used interchangeably. WIDGET-HANDLE is supported only for backward compatibility.

The following table describes the non-primitive (complex) types supported in ABL.

Table 22. ABL complex types
Complex type	Description
Array	An ABL array type is complex type that specifies a one-dimensional array of elements of the same scalar data type with a 1-based index, where a scalar data type is any data type that is not, itself, an array. The elements of an ABL array can contain scalars of any supported ABL primitive type, any ABL object type, or any .NET object type (other than a .NET mapped object type). The type definition for an ABL array is specified by the type definition for the array element with the addition of the EXTENT option (or with the Extent option selected in OpenEdge database tools). Thus, you can define an ABL array data element using similar features and syntax used to define primitive and object-type data elements. However, you cannot define ABL array types for the BLOB or CLOB primitive type. Also, note that unlike ABL arrays, .NET arrays are objects with an object type, just like any other .NET type. However, all .NET array object types derive from the System.Array class (a reference type), and you can create and access .NET arrays using public members of System.Array. Also, .NET arrays can be multi-dimensional and typically have a 0-based index. In .NET, the object type name of an array object is the object type name of its array elements appended with a set of square brackets ([]) with an embedded comma added for each additional dimension in the array. For example, System.Drawing.Point[] is a one-dimensional array object type and System.Drawing.Point[,] is a two-dimensional array object type. .NET languages support syntax for additional kinds of array objects, including jagged arrays. For more information on .NET array syntax, see the "Arrays Tutorial" in the C# Programmer's Reference on MSDN. In ABL, you must also enclose any .NET array object type name in double-quotes in order to handle the square brackets and any commas, which are special characters in ABL names, for example, "System.Drawing.Point[]". For more information on specifying array types, see the Type-name syntax reference entry.
Class	Classes encapsulate a set of data and behavioral elements (members), including implementations for data members, properties, and methods. Class definitions can also implement interfaces, which define a common set of prototypes for methods, properties, and events that classes can implement. For more information on classes and interfaces, see OpenEdge Development: Object-Oriented Programming.
Enumeration	An enumeration is made up of a list of strongly typed, named constants called members. The value of a variable defined as an enumeration type is restricted to the list of members defined for that enumeration type.
Handle-based object	A handle-based object has a built-in and inherent ABL type that provides data and behavior of varying complexity depending on the purpose of the object. A few examples include: Visual representation objects (widgets), such as buttons (defined using the DEFINE BUTTON statement, rectangles (defined using the DEFINE RECTANGLE statement statement), or data-representation widgets, such as a fill-ins (defined as part of the DEFINE VARIABLE statement or DEFINE PARAMETER statement) or browses (grid-like widgets defined using the DEFINE BROWSE statement) Data objects, such as temp-tables (defined using the DEFINE TEMP-TABLE statement), ProDataSets, (defined using the DEFINE DATASET statement), and related data objects, such as queries (defined using the DEFINE QUERY statement) Streams (defined using the DEFINE STREAM statement), which are used for reading and writing sequential data, such as text files Procedure objects (each defined as a file of ABL source code and created at run time using the RUN statement) Socket and server socket objects (created at run time using the CREATE SOCKET statement and CREATE SERVER-SOCKET statement Note: While each type of handle-based object is unique, because of their weak typing, you can reference all such objects using the same primitive type, HANDLE (see Table 21). For more information on each type of handle-based object, see the reference entry for its type in the Widget Reference or the Handle Reference, and see the reference entry for its respective DEFINE, CREATE, or other instantiating executable statement.

The following table lists the default data formats and initial values for ABL primitive and object types.

Table 23. Default ABL data type initial values and display formats
Data type	Default initial value	Default display format
BLOB¹, ²	Unknown value (?)	See footnote ³.
CHARACTER	"" (an empty string)	X(8)
CLASS⁴, ⁵,	Unknown value (?)	See footnote ⁶ .
CLOB⁷, ⁸	Unknown value (?)	See footnote ⁹.
COM-HANDLE¹⁰	Unknown value (?)	>>>>>>9
DATE	Unknown value (?) (displays as blanks)	99/99/99
DATETIME	Unknown value (?)	99/99/9999 HH:MM:SS.SSS
DATETIME-TZ	Unknown value (?)	99/99/9999 HH:MM:SS.SSS+HH:MM
DECIMAL	0	->>,>>9.99
HANDLE¹¹	Unknown value (?)	>>>>>>9
INT64	0	->,>>>,>>9
INTEGER	0	->,>>>,>>9
LOGICAL	no	yes/no
LONGCHAR¹²	"" (an empty string)	See footnote ¹³.
MEMPTR¹⁴, ¹⁵	A zero-length sequence of bytes	See footnote ¹⁶.
RAW¹⁷, ¹⁸	A zero-length sequence of bytes	See footnote ¹⁹.
RECID	Unknown value (?)	>>>>>>9
ROWID²⁰, ²¹	Unknown value (?)	See footnote ²².

¹ You cannot display a BLOB, CLOB, MEMPTR, RAW, or ROWID value directly. However, you can convert a MEMPTR, RAW, or ROWID value to a character string representation using the STRING function and display the result. You can also convert a BLOB to a MEMPTR, and then use the STRING function. A MEMPTR or RAW value converts to a decimal integer string. A ROWID value converts to a hexadecimal string, "0xhexdigits", where hexdigits is any number of characters "0" through "9" and "A" through "F". You can display a CLOB field by converting it to a LONGCHAR, and displaying the LONGCHAR using the VIEW-AS EDITOR LARGE phrase only.

² You cannot use the INITIAL option to specify an initial value for this data type as part of the definition of a variable, procedure parameter, or class-based property.

³ You cannot display a BLOB, CLOB, MEMPTR, RAW, or ROWID value directly. However, you can convert a MEMPTR, RAW, or ROWID value to a character string representation using the STRING function and display the result. You can also convert a BLOB to a MEMPTR, and then use the STRING function. A MEMPTR or RAW value converts to a decimal integer string. A ROWID value converts to a hexadecimal string, "0xhexdigits", where hexdigits is any number of characters "0" through "9" and "A" through "F". You can display a CLOB field by converting it to a LONGCHAR, and displaying the LONGCHAR using the VIEW-AS EDITOR LARGE phrase only.

⁴ If you display a class instance using the MESSAGE statement, ABL automatically invokes the ToString( ) method (Object) (provided by the Progress.Lang.Object class) on the object reference. To display a class instance in a frame (for example, using the DISPLAY statement), you must first explicitly convert the object reference to a displayable type using the INT64 function, the INTEGER function, the STRING function, or the ToString( ) method (Object).

⁵ You cannot use the INITIAL option to specify an initial value for this data type as part of the definition of a variable, procedure parameter, or class-based property.

⁶ You cannot display a BLOB, CLOB, MEMPTR, RAW, or ROWID value directly. However, you can convert a MEMPTR, RAW, or ROWID value to a character string representation using the STRING function and display the result. You can also convert a BLOB to a MEMPTR, and then use the STRING function. A MEMPTR or RAW value converts to a decimal integer string. A ROWID value converts to a hexadecimal string, "0xhexdigits", where hexdigits is any number of characters "0" through "9" and "A" through "F". You can display a CLOB field by converting it to a LONGCHAR, and displaying the LONGCHAR using the VIEW-AS EDITOR LARGE phrase only.

⁷ You cannot display a BLOB, CLOB, MEMPTR, RAW, or ROWID value directly. However, you can convert a MEMPTR, RAW, or ROWID value to a character string representation using the STRING function and display the result. You can also convert a BLOB to a MEMPTR, and then use the STRING function. A MEMPTR or RAW value converts to a decimal integer string. A ROWID value converts to a hexadecimal string, "0xhexdigits", where hexdigits is any number of characters "0" through "9" and "A" through "F". You can display a CLOB field by converting it to a LONGCHAR, and displaying the LONGCHAR using the VIEW-AS EDITOR LARGE phrase only.

⁸ You cannot use the INITIAL option to specify an initial value for this data type as part of the definition of a variable, procedure parameter, or class-based property.

⁹ You cannot display a BLOB, CLOB, MEMPTR, RAW, or ROWID value directly. However, you can convert a MEMPTR, RAW, or ROWID value to a character string representation using the STRING function and display the result. You can also convert a BLOB to a MEMPTR, and then use the STRING function. A MEMPTR or RAW value converts to a decimal integer string. A ROWID value converts to a hexadecimal string, "0xhexdigits", where hexdigits is any number of characters "0" through "9" and "A" through "F". You can display a CLOB field by converting it to a LONGCHAR, and displaying the LONGCHAR using the VIEW-AS EDITOR LARGE phrase only.

¹⁰ You cannot use the INITIAL option to specify an initial value for this data type as part of the definition of a variable, procedure parameter, or class-based property.

¹¹ You cannot use the INITIAL option to specify an initial value for this data type as part of the definition of a variable, procedure parameter, or class-based property.

¹² You cannot display a BLOB, CLOB, MEMPTR, RAW, or ROWID value directly. However, you can convert a MEMPTR, RAW, or ROWID value to a character string representation using the STRING function and display the result. You can also convert a BLOB to a MEMPTR, and then use the STRING function. A MEMPTR or RAW value converts to a decimal integer string. A ROWID value converts to a hexadecimal string, "0xhexdigits", where hexdigits is any number of characters "0" through "9" and "A" through "F". You can display a CLOB field by converting it to a LONGCHAR, and displaying the LONGCHAR using the VIEW-AS EDITOR LARGE phrase only.

¹³ You cannot display a BLOB, CLOB, MEMPTR, RAW, or ROWID value directly. However, you can convert a MEMPTR, RAW, or ROWID value to a character string representation using the STRING function and display the result. You can also convert a BLOB to a MEMPTR, and then use the STRING function. A MEMPTR or RAW value converts to a decimal integer string. A ROWID value converts to a hexadecimal string, "0xhexdigits", where hexdigits is any number of characters "0" through "9" and "A" through "F". You can display a CLOB field by converting it to a LONGCHAR, and displaying the LONGCHAR using the VIEW-AS EDITOR LARGE phrase only.

¹⁴ You cannot display a BLOB, CLOB, MEMPTR, RAW, or ROWID value directly. However, you can convert a MEMPTR, RAW, or ROWID value to a character string representation using the STRING function and display the result. You can also convert a BLOB to a MEMPTR, and then use the STRING function. A MEMPTR or RAW value converts to a decimal integer string. A ROWID value converts to a hexadecimal string, "0xhexdigits", where hexdigits is any number of characters "0" through "9" and "A" through "F". You can display a CLOB field by converting it to a LONGCHAR, and displaying the LONGCHAR using the VIEW-AS EDITOR LARGE phrase only.

¹⁵ You cannot use the INITIAL option to specify an initial value for this data type as part of the definition of a variable, procedure parameter, or class-based property.

¹⁶ You cannot display a BLOB, CLOB, MEMPTR, RAW, or ROWID value directly. However, you can convert a MEMPTR, RAW, or ROWID value to a character string representation using the STRING function and display the result. You can also convert a BLOB to a MEMPTR, and then use the STRING function. A MEMPTR or RAW value converts to a decimal integer string. A ROWID value converts to a hexadecimal string, "0xhexdigits", where hexdigits is any number of characters "0" through "9" and "A" through "F". You can display a CLOB field by converting it to a LONGCHAR, and displaying the LONGCHAR using the VIEW-AS EDITOR LARGE phrase only.

¹⁷ You cannot display a BLOB, CLOB, MEMPTR, RAW, or ROWID value directly. However, you can convert a MEMPTR, RAW, or ROWID value to a character string representation using the STRING function and display the result. You can also convert a BLOB to a MEMPTR, and then use the STRING function. A MEMPTR or RAW value converts to a decimal integer string. A ROWID value converts to a hexadecimal string, "0xhexdigits", where hexdigits is any number of characters "0" through "9" and "A" through "F". You can display a CLOB field by converting it to a LONGCHAR, and displaying the LONGCHAR using the VIEW-AS EDITOR LARGE phrase only.

¹⁸ You cannot use the INITIAL option to specify an initial value for this data type as part of the definition of a variable, procedure parameter, or class-based property.

¹⁹ You cannot display a BLOB, CLOB, MEMPTR, RAW, or ROWID value directly. However, you can convert a MEMPTR, RAW, or ROWID value to a character string representation using the STRING function and display the result. You can also convert a BLOB to a MEMPTR, and then use the STRING function. A MEMPTR or RAW value converts to a decimal integer string. A ROWID value converts to a hexadecimal string, "0xhexdigits", where hexdigits is any number of characters "0" through "9" and "A" through "F". You can display a CLOB field by converting it to a LONGCHAR, and displaying the LONGCHAR using the VIEW-AS EDITOR LARGE phrase only.

²⁰ You cannot display a BLOB, CLOB, MEMPTR, RAW, or ROWID value directly. However, you can convert a MEMPTR, RAW, or ROWID value to a character string representation using the STRING function and display the result. You can also convert a BLOB to a MEMPTR, and then use the STRING function. A MEMPTR or RAW value converts to a decimal integer string. A ROWID value converts to a hexadecimal string, "0xhexdigits", where hexdigits is any number of characters "0" through "9" and "A" through "F". You can display a CLOB field by converting it to a LONGCHAR, and displaying the LONGCHAR using the VIEW-AS EDITOR LARGE phrase only.

²¹ You cannot use the INITIAL option to specify an initial value for this data type as part of the definition of a variable, procedure parameter, or class-based property.

²² You cannot display a BLOB, CLOB, MEMPTR, RAW, or ROWID value directly. However, you can convert a MEMPTR, RAW, or ROWID value to a character string representation using the STRING function and display the result. You can also convert a BLOB to a MEMPTR, and then use the STRING function. A MEMPTR or RAW value converts to a decimal integer string. A ROWID value converts to a hexadecimal string, "0xhexdigits", where hexdigits is any number of characters "0" through "9" and "A" through "F". You can display a CLOB field by converting it to a LONGCHAR, and displaying the LONGCHAR using the VIEW-AS EDITOR LARGE phrase only.

For more information on using the built-in ABL primitive types, see OpenEdge Getting Started: ABL Essentials and the Web paper, ABL Data Types in OpenEdge Release 10.

As noted previously in this entry, ABL supports references to .NET types in two basic ways:

1. You can make direct and explicit reference to .NET object types using similar syntax that is supported for referencing ABL user-defined class and interface types. For supported .NET object types, this includes both the instantiation of a .NET class in ABL and the derivation of the .NET class by an ABL user-defined class (ABL-derived .NET class), and it includes the implementation of supported .NET interfaces by an ABL user-defined class (ABL-extended .NET class). To integrate the .NET class hierarchy with the ABL class hierarchy, ABL views System.Object as an immediate subclass of the ABL root class (Progress.Lang.Object class). In this way, you can manage .NET object types in ABL using many of the same mechanisms that you use for managing ABL class and interface types. However, you must observe the following limitations:

You cannot directly reference any .NET object type that is supported as a .NET mapped data type, except to define a .NET array of such types. You can only reference .NET mapped data types as their equivalent ABL built-in primitive types. For more information on .NET mapped data types, see the immediately following Step 2.

You cannot use System.Threading.Thread, or any derived class—ABL is single-threaded.

You cannot use System.MulticastDelegate, or any derived class (otherwise referred to as delegates) to provide handlers for .NET events. ABL provides its own event handling model for .NET events. For more information, see the ClassEvents Reference.

You cannot define an ABL interface that inherits from a .NET interface.

ABL imposes additional requirements on the .NET classes you can extend and the .NET interfaces you can implement in an ABL user-defined class. For more information, see the CLASS statement and INTERFACE statement reference entries.

For more information on the requirements for accessing .NET object types, see the notes section of this reference entry and OpenEdge Development: GUI for .NET Programming.

2. You can make implicit access to all .NET primitive types and their associated mapped object types by using the ABL built-in primitive types that are mapped to them. Because .NET mapped object types and .NET primitive types, together, represent the complete set of .NET mapped data types, the implicit mapping between .NET mapped data types and ABL primitive types allows you to access .NET method parameters, fields (data members), and properties using the corresponding ABL primitive types without direct reference to their .NET data type equivalents. In fact, ABL does not allow you to directly reference either the .NET primitive types or the .NET mapped object types as scalars without raising a compile-time error. (The exception is when defining a .NET array of mapped types. For more information, see the notes section of this reference entry.)

The following table shows the implicit mappings supported between .NET mapped data types and ABL built-in primitive types, showing the corresponding primitive types from C#.

Table 24. Implicit mappings between .NET and ABL data types
Implicit .NET object type	Implicit C# primitive type	ABL primitive type
System.Boolean	bool	LOGICAL
System.Byte	byte	INTEGER¹, ²
System.SByte	sbyte	INTEGER³
System.DateTime	N/A	DATETIME
System.Decimal	decimal	DECIMAL⁴, ⁵,
System.Int16	short	INTEGER⁶
System.UInt16	ushort	INTEGER⁷, ⁸
System.Int32	int	INTEGER⁹
System.UInt32	uint	INT64¹⁰, ¹¹
System.Int64	long	INT64¹²
System.UInt64	ulong	DECIMAL¹³, ¹⁴
System.Double	double	DECIMAL¹⁵
System.Single	float	DECIMAL¹⁶
System.Char	char	CHARACTER¹⁷
System.String	string	CHARACTER¹⁸ or LONGCHAR¹⁹, ²⁰

¹ An ABL INTEGER is a 32-bit number. Thus, it can hold values that are too big to store in a .NET System.Byte, System.SByte, System.Int16, or System.UInt16. Therefore, AVM raises a run-time error if an incompatible value is assigned.

² If you pass a negative ABL data type to an unsigned data type, the ABL virtual machine (AVM) raises a run-time error.

³ An ABL INTEGER is a 32-bit number. Thus, it can hold values that are too big to store in a .NET System.Byte, System.SByte, System.Int16, or System.UInt16. Therefore, AVM raises a run-time error if an incompatible value is assigned.

⁴ The range of values for a .NET System.Decimal and the range of values for an ABL DECIMAL are not equivalent. In particular, an ABL DECIMAL can be a much larger positive number or a much smaller negative number than a .NET System.Decimal can represent, and a .NET System.Decimal can represent a positive or negative number with much higher precision (with more significant digits to the right of the decimal point) than an ABL DECIMAL can represent. Therefore, the AVM raises a run-time error if you assign too large or too small of an ABL DECIMAL value to a .NET System.Decimal. If you assign too precise a .NET System.Decimal to an ABL DECIMAL, with too many significant digits to the right of the decimal point, ABL truncates the least significant digits necessary to represent the value as an ABL DECIMAL.

⁵ The .NET default match for this ABL primitive type when passed as an overloaded method parameter, when passed to a System.Object parameter, when used to define an overridden .NET method, when used to implement (or override) a .NET interface (or abstract) method, property, or event, or when converted using the BOX function, all without a specified AS data type indication. For more information on AS data types, see the following paragraphs and Table 25.

⁶ An ABL INTEGER is a 32-bit number. Thus, it can hold values that are too big to store in a .NET System.Byte, System.SByte, System.Int16, or System.UInt16. Therefore, AVM raises a run-time error if an incompatible value is assigned.

⁷ An ABL INTEGER is a 32-bit number. Thus, it can hold values that are too big to store in a .NET System.Byte, System.SByte, System.Int16, or System.UInt16. Therefore, AVM raises a run-time error if an incompatible value is assigned.

⁸ If you pass a negative ABL data type to an unsigned data type, the ABL virtual machine (AVM) raises a run-time error.

⁹ The .NET default match for this ABL primitive type when passed as an overloaded method parameter, when passed to a System.Object parameter, when used to define an overridden .NET method, when used to implement (or override) a .NET interface (or abstract) method, property, or event, or when converted using the BOX function, all without a specified AS data type indication. For more information on AS data types, see the following paragraphs and Table 25.

¹⁰ An ABL INT64 is a 64-bit number. Thus, it can hold values that are too big to store in a .NET System.UInt32. Therefore, AVM raises a run-time error if an incompatible value is assigned.

¹¹ If you pass a negative ABL data type to an unsigned data type, the ABL virtual machine (AVM) raises a run-time error.

¹² The .NET default match for this ABL primitive type when passed as an overloaded method parameter, when passed to a System.Object parameter, when used to define an overridden .NET method, when used to implement (or override) a .NET interface (or abstract) method, property, or event, or when converted using the BOX function, all without a specified AS data type indication. For more information on AS data types, see the following paragraphs and Table 25.

¹³ If you pass a negative ABL data type to an unsigned data type, the ABL virtual machine (AVM) raises a run-time error.

¹⁴ An ABL DECIMAL can represent a much larger number than a System.UInt64. Therefore, AVM raises a run-time error if an incompatible value is assigned.

¹⁵ An ABL DECIMAL represents numbers up to 50 digits long. As a result, an ABL DECIMAL value cannot represent the full range of values for a .NET System.Double or System.Single. Therefore, AVM raises a run-time error if an incompatible value is assigned. Also, an ABL DECIMAL can lose precision when it is represented by a .NET System.Double or System.Single.

¹⁶ An ABL DECIMAL represents numbers up to 50 digits long. As a result, an ABL DECIMAL value cannot represent the full range of values for a .NET System.Double or System.Single. Therefore, AVM raises a run-time error if an incompatible value is assigned. Also, an ABL DECIMAL can lose precision when it is represented by a .NET System.Double or System.Single.

¹⁷ This ABL CHARACTER mapping supports a single Unicode character.

¹⁸ The .NET default match for this ABL primitive type when passed as an overloaded method parameter, when passed to a System.Object parameter, when used to define an overridden .NET method, when used to implement (or override) a .NET interface (or abstract) method, property, or event, or when converted using the BOX function, all without a specified AS data type indication. For more information on AS data types, see the following paragraphs and Table 25.

¹⁹ The .NET default match for this ABL primitive type when passed as an overloaded method parameter, when passed to a System.Object parameter, when used to define an overridden .NET method, when used to implement (or override) a .NET interface (or abstract) method, property, or event, or when converted using the BOX function, all without a specified AS data type indication. For more information on AS data types, see the following paragraphs and Table 25.

²⁰ When the value comes from .NET, ABL converts from System.String to either CHARACTER or LONGCHAR, depending on the length of the character string. When the value comes from ABL, System.String accepts values from either CHARACTER or LONGCHAR.

Thus, instead of using an object reference to the corresponding .NET mapped object type, you must provide or access all .NET primitive (or mapped object type) values for .NET methods, data members, and properties as ABL primitive types. Similarly, when you reference any data element or value defined as a .NET mapped data type, ABL evaluates the .NET value to its corresponding ABL primitive value. ABL checks for .NET/ABL type compatibility at compile time, except in rare cases where data type narrowing is allowed, in which case the AVM checks for data overflow or underflow at run time.

Note: To access all other .NET value types in ABL except mapped data types (for example, System.Drawing.Size), you can and must use object references to the value type objects.

.NET supports a concept known as boxing. Boxing is the process of converting a value type (such as a C# int or .NET System.Int32) to a reference type object. Boxing a value type wraps its value inside a System.Object. Unboxing extracts the value from the System.Object as the original value type. In .NET, boxing and unboxing between a value type and a System.Object occurs during assignment or parameter passing.

So, in addition to implicitly mapping its native primitive types to their corresponding .NET mapped data types, ABL also supports boxing between its primitive or array types and a .NET System.Object or array object. ABL performs boxing operations automatically in two cases:

When you assign values between a .NET System.Object or one-dimensional array object and a compatible ABL primitive or array type

When you pass parameter values for .NET methods and constructors between a .NET System.Object or one-dimensional array object and a compatible ABL primitive or array type

However, as described further in this entry, ABL does not support automatic boxing operations when passing parameters to ABL routines.

When ABL does automatic boxing that involves ABL primitive types or arrays of elements containing primitive types, it also does implicit conversion between these types and the corresponding .NET mapped types (see Table 24). For example, if you assign an ABL INTEGER to a System.Object, ABL converts the ABL INTEGER to a System.Int32, which the System.Object accepts as a subclass value. Similarly, if you assign an ABL INTEGER array to a System.Object, ABL converts the ABL INTEGER array to a "System.Int32[]", which the System.Object accepts as a subclass value. The same occurs when you pass an ABL INTEGER or INTEGER array to a System.Object INPUT parameter of a .NET method.

In reverse, when you assign an appropriate System.Object to an ABL INTEGER or INTEGER array, ABL unboxes the System.Object by determining the .NET mapped type that the System.Object represents, converts that value to its equivalent ABL primitive or primitive array value, and attempts to assign the result to the ABL INTEGER or INTEGER array (which is validated at run time). For example, if the System.Object represents the System.Decimal subclass and you are assigning it to an ABL INTEGER, ABL converts the System.Decimal value to an ABL DECIMAL and attempts to assign it to the ABL INTEGER.

In a similar manner, ABL also does automatic boxing directly between compatible ABL arrays and one-dimensional .NET array objects. For example, if you assign or pass .NET method parameters between a "System.Windows.Forms.Button[]" array object and an ABL array of System.Windows.Forms.Button elements, ABL automatically does the required boxing and unboxing to convert between the different array types. A similar boxing and unboxing operation occurs between an ABL primitive array and a compatible .NET array of mapped types, for example, between an ABL array of INTEGER and a .NET "System.Int16[]" array object. For more information on boxing and unboxing between ABL and .NET arrays, see the notes section in this reference entry on working with .NET arrays.

However in the following four ABL contexts, automatic ABL boxing or unboxing is either not supported or might not be supported as you require:

When you use a System.Object directly in an expression, ABL does not unbox the System.Object into a compatible ABL primitive type.

When you assign an ABL primitive value (or primitive array) to a System.Object and the ABL primitive type maps to multiple .NET data types, it automatically boxes the ABL primitive value (or primitive array elements) as the default matching .NET mapped object type, which might not be the .NET data type mapping that you require.

ABL does no automatic boxing or unboxing when you pass an ABL primitive or array type to a compatible .NET object parameter of an ABL routine (ABL method, procedure, or user-defined function). Similarly, ABL also does no automatic boxing or unboxing when you pass a compatible .NET object argument to the ABL primitive or array parameter of an ABL routine.

ABL does no boxing or unboxing of array elements when you box or unbox an ABL array. For example, ABL cannot assign between a "System.Object[]" and an ABL INTEGER array, because it does not handle boxing and unboxing between the corresponding System.Object and the INTEGER array elements. For more information on ABL support for array assignments, see the notes section of this reference entry.

When you use a System.Object directly in an expression, ABL raises a compile-time error because ABL does not support automatic unboxing of a System.Object in an expression. Instead, you can use the ABL built-in UNBOX function in the expression to explicitly unbox the value. This function accepts the System.Object as input and returns an ABL primitive value that is equivalent to the .NET mapped object type value (subclass) represented by the specified .NET System.Object instance.

When you assign an ABL primitive value or primitive array to a System.Object, ABL always boxes the value or array into a particular .NET mapped type or array of mapped types, which might not be the .NET type you want. In Table 24, several ABL primitive types implicitly map to more than one .NET mapped data type. For each ABL primitive type that maps to multiple .NET data types, ABL uses one of these mappings as the .NETdefault match for the ABL primitive type (indicated by a footnote ¹ in Table 24). Thus, when you assign an ABL primitive value or primitive array to a System.Object, ABL automatically boxes the value or array using the .NET default match for the specified ABL data type. For example, by default an INTEGER automatically boxes as a System.Int32, and an INTEGER array automatically boxes as a "System.Int32[]".

However, you can explicitly box the ABL value or array using a .NET mapped type other than the default match with the ABL built-in BOX function. This function accepts an ABL primitive value or array as input and, by default, returns a boxed .NET type according to the .NET default match for the ABL data type of the input value or array. In order to box the value using a mapped type other than the .NET default match, you can pass an ABL keyword as a string to the function that indicates the explicit .NET type you want to use. For example, an ABL DECIMAL value can represent both a .NET System.Decimal (the default match) and a System.Double (among other possible types). If you need to box the ABL DECIMAL as a .NET System.Double, you can explicitly indicate this to the BOX function. Similarly, if you need to box an ABL DECIMAL array as a "System.Double[]", you can use the same indication.

Note: When you unbox a System.Object using the UNBOX function, you cannot similarly specify a particular ABL primitive or primitive array type as the result. ABL always unboxes any System.Object using the default matching ABL type.

Another case for which you must use the BOX function or the UNBOX function is when you pass parameters between compatible ABL primitive or array types and .NET object types in the parameters of ABL methods, procedures, and user-defined functions. ABL raises a compiler error if you try to pass these types to each other directly in ABL routine parameters. Appropriate use of the BOX function or UNBOX function allows this type of parameter passing to occur without a compile-time error. Note, again, that ABL does support the automatic boxing and unboxing of .NET objects in parameter passing for .NET method calls.

Similarly, three additional cases exist (other than the need for explicit boxing) where you must specify the .NET data type mapping you want for a given ABL primitive type:

When a method parameter is overloaded by multiple implicit .NET data type mappings for a passed ABL primitive type. You must specify the exact .NET data type when you pass the parameter to the method.

When you override a method inherited from a .NET class, or when you implement (or override) a .NET interface (or abstract) method, property, or event, and the types of any associated parameters, properties, or return values are .NET mapped data types. You must specify the exact .NET data type in the definition of each .NET mapped parameter, property, and return value.

When you pass an ABL primitive value (not an array) to a .NET method or constructor parameter that is a System.Object, and you want the result to be a .NET mapped type other than the default match. You must indicate the explicit mapped type on the passed ABL primitive argument. A common use case is the SetValue( ) method of the System.Array class, which sets the value of a .NET array element. If the .NET type of the array element is other than the default match, you must indicate the .NET mapped type for the value parameter to match the array definition.

When you reference a constructed .NET generic type (described further in this entry) using type parameters that include a .NET mapped type.

To indicate a non-default .NET mapped type in the previous cases where you want an explicit mapped type to be used, the syntax for the following ABL elements allows you to specify an appropriate ABL keyword:

BOX function

DEFINE PROPERTY statement

METHOD statement

Parameter definition syntax

Parameter passing syntax

This keyword is referred to as an AS data type, because you specify it for a passed parameter using the AS option. So, for example, when you override a .NET method, you must explicitly specify .NET data type for each .NET mapped parameter, property, or return type. If the .NET data type is a default match, you must simply use the matching ABL data type. Otherwise, you must indicate the appropriate AS data type keyword for the data type of the method parameter, return type, or property definition.

Table 25 lists each explicit .NET data type mapping for a given ABL primitive type. For each listed .NET data type, you indicate this explicit mapping either by using the corresponding ABL primitive type (for a default match) or by using the appropriate option to specify the AS data type that corresponds to the explicit .NET data type you want to map. For more information on specifying the AS data type option when using the BOX function or when calling overloaded .NET methods, see the reference entries for the BOX function and Parameter passing syntax in this book. For more information on specifying AS data types when overriding a .NET method, or when implementing (or overriding) a .NET interface (or abstract) method, property, or event, see the METHOD statement, the DEFINE PROPERTY statement, the DEFINE EVENT statement, or the Parameter definition syntax reference entry, as appropriate.

Note: The AS data types in Table 25 represent some different data types than you can specify using the AS option to pass a COM method parameter. For more information on passing COM method parameters, see Syntax for accessing COM object properties and methods.

Table 25. Explicit mappings between ABL and .NET data types
Explicit .NET object type	Explicit C# primitive type	ABL primitive type	ABL AS data type
System.Boolean	bool	LOGICAL¹	–
System.Byte	byte	INTEGER	UNSIGNED-BYTE²
System.SByte	sbyte	INTEGER	BYTE
System.DateTime	N/A	DATETIME³	–
System.Decimal	decimal	DECIMAL⁴	–
System.Int16	short	INTEGER	SHORT
System.UInt16	ushort	INTEGER	UNSIGNED-SHORT
System.Int32	int	INTEGER⁵	–
System.UInt32	uint	INT64	UNSIGNED-INTEGER
System.Int64	long	INT64⁶	–
System.UInt64	ulong	DECIMAL	UNSIGNED-INT64
System.Double	double	DECIMAL	DOUBLE
System.Single	float	DECIMAL	FLOAT
System.Char	char	CHARACTER	SINGLE-CHARACTER
System.String	string	CHARACTER/ LONGCHAR⁷	–

¹ The explicit .NET data type is the default match for the corresponding listed ABL primitive type. To specify a .NET default matching data type, simply pass the ABL primitive value as an overloaded method parameter or BOX function parameter, or define the overridden or implemented .NET method parameter, return type, or property as the corresponding ABL primitive type.

² To disambiguate overloaded methods, you cannot pass an ABL array and use an AS data type on an array parameter of a .NET method call. You must pass the argument as the .NET-defined array object type instead.

³ The explicit .NET data type is the default match for the corresponding listed ABL primitive type. To specify a .NET default matching data type, simply pass the ABL primitive value as an overloaded method parameter or BOX function parameter, or define the overridden or implemented .NET method parameter, return type, or property as the corresponding ABL primitive type.

⁴ The explicit .NET data type is the default match for the corresponding listed ABL primitive type. To specify a .NET default matching data type, simply pass the ABL primitive value as an overloaded method parameter or BOX function parameter, or define the overridden or implemented .NET method parameter, return type, or property as the corresponding ABL primitive type.

⁵ The explicit .NET data type is the default match for the corresponding listed ABL primitive type. To specify a .NET default matching data type, simply pass the ABL primitive value as an overloaded method parameter or BOX function parameter, or define the overridden or implemented .NET method parameter, return type, or property as the corresponding ABL primitive type.

⁶ The explicit .NET data type is the default match for the corresponding listed ABL primitive type. To specify a .NET default matching data type, simply pass the ABL primitive value as an overloaded method parameter or BOX function parameter, or define the overridden or implemented .NET method parameter, return type, or property as the corresponding ABL primitive type.

⁷ The explicit .NET data type is the default match for the corresponding listed ABL primitive type. To specify a .NET default matching data type, simply pass the ABL primitive value as an overloaded method parameter or BOX function parameter, or define the overridden or implemented .NET method parameter, return type, or property as the corresponding ABL primitive type.

A .NET generic type is a class or interface defined so that it functions as one of several different types, depending on how you reference its type name. A reference to a .NET generic type name includes one or more type parameters, each of which specifies a data type that the generic type can use in its implementation. When you reference the generic type name in ABL, you substitute a specific data type for each type parameter defined for the generic type. This reference then identifies the generic type as a constructed type. The notation for a generic type that you see in .NET documentation or in a class browser, where the type parameters are not resolved, is called an open type. An open type reference contains only placeholders for the parameters in the type name, such as <T>, which defines the single parameter for the following generic type:

System.Collections.Generic.List<T>

In ABL, you can only reference a .NET generic type as a constructed type using a type name that has the following syntax:

"namespace.object-name<type-parameter[ , type-parameter ]...>"

The namespace is a .NET namespace and the syntax from object-name up to and including the right angle bracket (>) forms the dotNET-object-name as described in the Type-name syntax reference entry. The left and right angle brackets (<>) are the part of .NET generic type name references that enclose the type parameters, as shown in the previous example. Each type-parameter in the parameter list represents a placeholder for a specific .NET data type. The number of type parameters and the data type that you can specify for each type-parameter in a constructed type reference depends on the generic type definition. However, you can never specify an ABL object type or an ABL-extended .NET class type as the data type of any type-parameter; it can only be a pure .NET type. The quotes are required in order to allow for the angle brackets and any spaces in the type name.

The definition for each type-parameter in a .NET generic type definition can specify constraints that determine the .NET types you can substitute for a given parameter when you reference the constructed type. If these constraints on a type-parameter allow you to specify one or more .NET mapped types, you must specify an appropriate explicit mapping for each such type when you specify the type-parameter in ABL, as described in Table 25.

For example, to define an object reference to a System.Collections.Generic.List<T> that is constructed as a list of System.Int16, you might use the following ABL statement:

DEFINE VARIABLE shortList AS
CLASS System.Collections.Generic.List<SHORT> NO-UNDO.

You can also reference an array of a generic type and define generic types with a type parameter that is an array. For more information, see the information on .NET arrays in the notes of this reference entry.

You can use a .NET generic type in all the same contexts as any other .NET type except to define an ABL class that:

Inherits from a .NET generic class

Implements a .NET generic interface

Also, while you can cast an object reference to a .NET generic type using the CAST function, you cannot cast to a .NET generic type using the DYNAMIC-CAST function.

For more information on how to identify .NET generic types and understand the constraints on their type parameters, see the .NET documentation on MSDN. For more information on working with .NET generic types in ABL, see OpenEdge Development: GUI for .NET Programming.

Notes

ABL provides built-in data types, built-in class data types, and user-defined class data types. References to built-in data types follow these rules:

Like most other keywords, references to specific built-in data types appear in all UPPERCASE, using a font that is appropriate to the context. No uppercase reference ever includes or implies any data type other than itself.

Wherever integer appears, this is a reference to the INTEGER or INT64 data type.

Wherever character appears, this is a reference to the CHARACTER, LONGCHAR, or CLOB data type.

Wherever decimal appears, this is a reference to the DECIMAL data type.

Wherever numeric appears, this is a reference to the INTEGER, INT64, or DECIMAL data type.

References to built-in class data types appear in mixed case with initial caps, for example, Progress.Lang.Object. References to user-defined class data types appear in mixed case, as defined for a given application example.

INT64 support applies to all of the ABL built-in methods and functions that take integer-expression parameters. That is, integer-expression parameters can be either INT64 expressions or INTEGER expressions.

Starting with Release 10.1B, all intermediate calculations are carried out in 64-bit arithmetic. For example, 2,000,000,000 * 100 / 100 gives the correct result whether the target field is INTEGER or INT64. However, although 2,000,000,000 * 100 does not cause an overflow, you must assign the result to an INT64 field. If you assign it to an INTEGER field, the AVM generates a run-time error.

In Version 9.0, when you copy one MEMPTR (M1) to another MEMPTR (M2), only the MEMPTR address is copied and both MEMPTRs point to the same memory location (L1). You can change the data in the single memory location and both MEMPTRs will point to the changed data. To clear memory after using the MEMPTRs, you can SET-SIZE = 0 on just one of the MEMPTRs.

Starting with Version 9.1, when you copy one MEMPTR (M1) to another MEMPTR (M2), the data that M1 points to is also copied. Therefore, MEMPTR M1 points to memory location L1, and MEMPTR M2 now points to memory location L2 which contains a copy of the data in L1. You must change the data in both memory locations if you want both MEMPTRs to reflect the change. To clear memory after using the MEMPTRs, you must execute SET-SIZE = 0 on both MEMPTRs to be sure that both memory locations are cleared.

Since RAW variables are limited in size to 32K and MEMPTR variables are not limited in size, if a MEMPTR with a size greater than 32K is copied to a RAW variable, the AVM generates an error.

Both a primitive type or object type can be defined as an array. Use the EXTENT option when defining or creating a field, variable, or object to establish an array. For example:

DEFINE VARIABLE someIntArray AS INTEGER EXTENT 4.

The variable someIntArray is now defined as an array of four integers. Since the size is fixed at 4, this is a determinate array. You can also define an indeterminate array by omitting the constant integer value after EXTENT. In this case, the number of elements in the array is undefined.

To refer to an individual element in an array, enclose the INTEGER index (subscript) using bracket syntax. This is known as a subscripted array reference. For example:

ASSIGN someIntArray[ 2 ] = 128.

Here, 2 references the second element in the INTEGER array.

Arrays can also be manipulated as a whole for array-to-array deep copy operations and to pass or return parameters. By omitting the brackets, a reference to the field, variable, or object name is a reference to the entire array. This is called an unsubscripted array reference. For example:

ASSIGN someIntArray = anotherIntArray.

Here, each element of the anotherIntArray will be copied into the corresponding element of the someIntArray. This is called a deep copy. Note that unsubscripted array references are not supported in expressions or comparison operations. For more information on array assignments, see the Assignment (=) statement reference entry.

ABL supports access to the following kinds of .NET object types:

Classes — Viewed and managed like ABL classes with support for additional features that are unique to .NET classes, such as inner classes and indexers for indexed properties. For information on accessing instances of .NET classes, see the Class-based object reference entry.

Interfaces — Viewed and managed as ABL interfaces with support for additional features that are unique to .NET, such as inner interfaces.

Structures — Viewed and managed similar to ABL classes. Structure types are supported using syntax native to each .NET language, for example, using the struct keyword in C# and C++. For information on accessing instances of .NET structures, see the Class-based object reference entry. The essential difference between .NET structures and most other .NET classes is that structures inherit from System.ValueType and are therefore value types. Thus, all structures are passed within .NET, and between ABL and .NET, by value. However within ABL, structure objects are passed, like all other class instances, by reference. Therefore, when you access a structure from .NET, you reference a copy of the object in ABL that is separate from the object in .NET, and when you pass an ABL reference to a structure back to .NET, .NET gets a copy of the object that is separate from the object that is referenced in ABL. Structures therefore have different object management requirements in ABL than reference type objects. For more information, see the information on ABL support for value types in OpenEdge Development: GUI for .NET Programming.

Enumerations — Like ABL enumerations, .NET enumerations are classes that correspond to a named set of constant values with a single underlying data type. Each of these constant values corresponds to a member of a given enumeration class. Each .NET language allows you to define and reference enumeration members using its own syntax. ABL also provides syntax that allows you to reference .NET enumerations as object types. For more information, see the Enumeration member access reference entry. Enumerations inherit from the System.Enum structure, which inherits from System.ValueType. Thus, like structures, enumerations are value types that are passed by value between .NET and ABL. However, unlike .NET languages that can view enumerations as values, ABL views enumerations only as objects that are passed by reference, like any other class instance. Enumerations therefore have similar object management requirements to structures in ABL. For more information, see the information on ABL support for value types in OpenEdge Development: GUI for .NET Programming.

You can instantiate .NET class or structure instances using the NEW function (classes), as with any ABL class. However, you cannot create an enumeration object. ABL can only reference enumeration objects that have already been defined in .NET. For information on specific .NET object types, see the documentation provided by the vendor for that object type.

ABL supports widening relationships between certain ABL data types. Widening allows you to pass an argument to a method parameter that has a different data type than the parameter, depending on the data flow (INPUT or OUTPUT). Thus, the target of the data flow can be a different data type if it can hold the largest value provided by the source of the data flow. When passing .NET method parameters or getting and setting .NET property values, ABL supports additional widening relationships between the ABL data type being passed and the .NET data type of the parameter. For more information, see the description of widening for .NET parameters in the Parameter passing syntax reference entry.

A .NET array is an object that extends the System.Array class. You can access a .NET array in ABL using an object reference, like any other object. Thus, in ABL, you can access all .NET arrays, of all dimensions, whose element type is either an ABL-supported .NET object type (such as System.Windows.Forms.Form) or a .NET mapped data type (such as System.Int16 or C# short). You can also create .NET arrays directly in ABL by creating instances of the System.Array class. Note that while you cannot explicitly define a variable as a System.Int16, ABL does allow you to define a .NET array where the element type is a System.Int16 (or any other mapped object type). .NET arrays have the following class hierarchy in ABL, in order of derivation from the ABL root class:

1. Progress.Lang.Object

2. System.Object

3. System.Array

4. Any array class of a specified element type, for example, "System.Int32[]", which specifies a one-dimensional array of System.Int32 elements

You can define references to .NET arrays or ABL arrays (extents) of .NET array references:

DEFINE VARIABLE buttonArray AS
CLASS "System.Windows.Forms.Button[]" NO-UNDO.

DEFINE VARIABLE buttonArrayExt AS
CLASS "System.Windows.Forms.Button[]" EXTENT 3 NO-UNDO.

You can also define references to arrays of a generic type or to generic types that have type parameters that are arrays. For example:

DEFINE VARIABLE shortListArrayObj AS
  CLASS "System.Collections.Generic.List<SHORT>[]" NO-UNDO.

DEFINE VARIABLE shortListArrayExt AS
  CLASS "System.Collections.Generic.List<SHORT>" EXTENT 3 NO-UNDO.

DEFINE VARIABLE buttonArrayList AS CLASS
  "System.Collections.Generic.List<System.Windows.Forms.Button[]>"
  NO-UNDO.

DEFINE VARIABLE buttonArrayListExt AS CLASS
  "System.Collections.Generic.List<System.Windows.Forms.Button[]>"
  EXTENT 3 NO-UNDO.

Because all .NET array objects inherit from System.Array, you can access any .NET array object using the members of the System.Array class. To help create .NET array objects, OpenEdge provides a Progress.Util.TypeHelper class to specify the System.Type object needed for creating .NET array objects. For more information on working with .NET arrays, see the sections on accessing .NET arrays in OpenEdge Development: GUI for .NET Programming.

If a .NET array is a multi-dimensional array, you can only work with it as a .NET array object, using the System.Array access mechanisms. However, if a .NET array is a one-dimensional array, you can work with it in two different ways:

Directly as a .NET array object

As an ABL array by directly assigning the .NET array to an equivalent ABL array, working with the resulting ABL array using ABL mechanisms, and directly assigning the reworked ABL array back to a .NET array

An array assignment can occur between .NET and ABL arrays of compatible element types, either by direct assignment of one array to another using the Assignment (=) statement or the ASSIGN statement, or by passing array parameters to .NET methods using Parameter passing syntax. In these specific cases, ABL performs automatic boxing and unboxing between the compatible ABL and .NET array types. In general, you can assign an ABL or .NET array of .NET value types (such as System.Drawing.Size) only to another ABL or .NET array of identical value type elements. ABL makes an exception if the ABL array in the assignment is an array of primitive type elements (such as INTEGER), in which case ABL allows assignment to or from a compatible .NET array of mapped types (such as "System.Int32[]" or "System.Byte[]"). Otherwise (for reference types), elements of the target array must be identical to or higher in the class hierarchy than the elements of the source array. For example, you can assign an array of System.Windows.Forms.Form elements to an array of System.Object elements.

While ABL does automatically box and unbox entire arrays for supported array assignments, ABL does not automatically box and unbox the elements of the source and target arrays. So, for example, you cannot assign an ABL array of INTEGER elements to a .NET array of System.Object elements ("System.Object[]").

Given that array element types are compatible, how an array assignment works, depends on the system context (ABL or .NET) of the arrays involved. In general, if either or both the target or source of the assignment is an ABL array, this results in a deep copy of all the array elements from the source array to the target. If both the target and source of the assignment is a .NET array, as with any object assignment, this results in an object reference copy, where the target references the same array as the source. Other array interactions depend on the array type of the target.

Caution: Because assignment between a .NET array and an ABL array requires a deep copy, note the performance impact it might have on your application before coding this operation. For a .NET array, you might prefer to work directly with the object reference, and access individual elements using System.Array mechanisms.

For a .NET and ABL array assignment where the target is an ABL array:

The source array must be another ABL array, a one-dimensional .NET array, or a System.Object whose type at run time is a one-dimensional .NET array, and the element data type of the target array must be compatible with the element data type of the source, or the assignment raises an error. In general, the compatibility between source and target element types follows standard ABL rules for data type assignments, both within ABL and between ABL and .NET. For more information on data type compatibility for assignment and parameter passing, see the Assignment (=) statement and the Parameter passing syntax reference entry.

If the target is an ABL array of System.Object elements and the source array is an array of .NET mapped types (such as System.Int32) or ABL primitive types, ABL raises a compile-time error, because (as noted previously) ABL does not automatically do the unboxing and boxing operations that are required on the elements of each array.

If the target has an indeterminate EXTENT, the elements from the source array are copied to the target, fixing its EXTENT to the number of elements in the source array.

If the target has a fixed EXTENT and the number of elements in the source array match the target EXTENT, the elements from the source array are copied to the target; otherwise, the assignment raises an error. This is true for both ABL and .NET source arrays.

For a .NET and ABL array assignment where the target is a .NET array object or other object reference:

The source array can be any ABL array or .NET array with an element data type that is compatible with the target. If the element data type of the target array is a .NET value type, a .NET source array must be defined with elements of an identical value type. If the target array is a .NET array of mapped types, an ABL source array must be defined with elements of an ABL primitive type that implicitly maps to the target element type (see Table 24. If the target element type is not mapped (such as System.Drawing.Size), an ABL source array (like a .NET array) must be defined with elements of an identical value type. Otherwise, assignment to arrays of .NET reference type elements follow standard rules for assigning object references of related class and interface types.

If the target is a System.Array or a System.Object, you can assign to it any .NET source array object or any ABL source array that is defined as a supported .NET object type or as an ABL primitive type that maps implicitly to a .NET mapped data type (see Table 24). If the source is a .NET array object, ABL simply assigns it to the System.Array or System.Object reference. If the source is an ABL array, ABL creates a new .NET array object to hold the ABL array elements and assigns it to the System.Array or System.Object reference. In addition, if the ABL array elements have a primitive type, ABL automatically maps the ABL array elements into the default matching .NET object type before storing them in the specified System.Array or System.Object.

If the target is a .NET array of System.Object elements ("System.Object[]"), you can also assign any compatible ABL or .NET array to it. Note (as previously described) that a source array with .NET value type elements (such as System.Drawing.Size or ABL INTEGER, which resolves to the value-type, System.Int32) is not compatible. .NET requires that the element types must be identical in array assignments involving value type elements.

If the target is a Progress.Lang.Object, you can assign to it any .NET source array object, but not a native ABL array (which is not an object).

The EXTENT of any ABL source array and the dimensions and size of a .NET source array do not matter. As noted previously, ABL copies the elements of an ABL source array to a newly created .NET array object and stores the object into the target object reference.

Caution: The index for ABL arrays is 1-based, while the index for .NET arrays is generally 0-based.

You can pass the Unknown value (?) as a parameter to a .NET method or assign the Unknown value (?) to a .NET property or data member. ABL translates the Unknown value (?) in these cases to the .NET null value. For the numeric and logical .NET primitive types listed in Table 24, when they are set to null, .NET sets a different default value—0, 0.0, or no—depending on the data type. In these cases, ABL also returns the .NET null value as the ABL Unknown value (?).

ABL does not do any mapping between System.Data.DataSet or System.Data.DataTable method parameters, properties, or data members on one hand and ABL ProDataSets and temp-tables on the other. ABL supports data binding between ProDataSets or temp-tables (among other data sources) and .NET form controls using the Progress.Data.BindingSource class (the ProBindingSource). For more information, see the Progress.Data.BindingSource class reference entry. However, you can always directly access .NET DataSet and DataTable objects as any other .NET object, using their class members.

Data types

Notes

See also