.NET Predefined Native Types
COBOL provides predefined USAGES corresponding to many of the most commonly used .NET types. These names can be used in the USAGE phrase when declaring a data item, and anywhere that a classname is expected.
| Predefined COBOL USAGE | .NET Class | C# equivalent | Alternative COBOL |
|---|---|---|---|
| binary-char | System.SByte | sbyte | PIC S9(2) COMP-5 |
| binary-char unsigned | System.Byte | byte | PIC 9(2) COMP-5 |
| binary-short | System.Int16 | short | PIC S9(4) COMP-5 |
| binary-short unsigned | System.UInt16 | ushort | PIC 9(4) COMP-5 |
| binary-long | System.Int32 | int | PIC S9(9) COMP-5 |
| binary-long unsigned | System.UInt32 | uint | PIC 9(9) COMP-5 |
| binary-double | System.Int64 | long | PIC S9(18) COMP-5 |
| binary-double unsigned | System.UInt64 | ulong | PIC 9(18) COMP-5 |
| character | System.Char | char | |
| float-short | System.Single | float | USAGE COMP-1 |
| float-long | System.Double | double | USAGE COMP-2 |
| condition-value | System.Boolean | bool | |
| decimal | System.Decimal | decimal | |
| object | System.Object | object | USAGE OBJECT REFERENCE |
| string | System.String | string |
Further .NET types can be specified in COBOL using the syntax USAGE class-name-1 or USAGE OBJECT REFERENCE class-name-1. All such data items and any of USAGE DECIMAL, OBJECT or STRING must be:
- Declared at 01 (or 77) level
- Have no Redefines or Renames
- Never be the subject of reference modification
- Never be the subject of the ADDRESS OF clause
They may make use of Format 3 of the OCCURS clause in order to declare a .NET array.
Furthermore, in order to be mapped onto the corresponding .NET native type, any of the other predefined USAGEs listed above must also follow these rules.
Any COBOL data item that does not follow these rules, or is of any other category, is not considered to be a .NET native type and is allocated by the compiler with an internally managed byte array. COBOL pointer data items always point into one of these byte arrays.
.NET distinguishes two types of objects:
- Value types (e.g. System.Int32, System.Decimal, System.DateTime). For data of this type, the data item you define in your program contains the actual data. E.g. for a System.Int32 (BINARY-LONG in COBOL), the data item will contain a 32 bit integer.
- Reference types, which are allocated on the object heap. For this type of data, the field in your program holds a reference into the object heap. Use of Reference types causes overhead due to indirection and the need to tidy up the heap (Garbage Collection).
All value types can be turned into reference types by a process known as boxing. This happens automatically when needed, for example when a value type is passed as a parameter to a method that expects a System.Object as a parameter. It can be done explicitly by assigning a value type to an item of type System.Object (i.e. a generic object). Boxed value types can be unboxed to restore the original value type.
When you specify USAGE [OBJECT REFERNCE] class-name:
- If class-name represents a reference type, you get a reference type object.
- If class-name represents a value type, you get a value type object.
.NET arrays are defined with the format 3 OCCURS clause.