Software AG IDL to C Mapping

This document describes the specific mapping of Software AG IDL data types, groups, arrays and structures to the C programming language. See also hints and restrictions on the Software AG IDL data types valid for all programming language bindings under IDL Data Types in the IDL Editor documentation.

This document covers the following topics:


Mapping IDL Data Types to C Data Types

In the table below, the following metasymbols and informal terms are used for the IDL.

  • The metasymbols "[" and "]" enclose optional lexical entities.

  • The informal term number (or in some cases number1.number2) is a sequence of numeric characters, for example 123.

Software AG IDL Description C Data Type Note
Anumber Alphanumeric unsigned char [number] 1,3
char [number + 1] 1,3
AV Alphanumeric variable length ERX_HVDATA 4
AVnumber Alphanumeric variable length with maximum length ERX_HVDATA 3,4
Bnumber Binary unsigned char [number] 3
BV Binary variable length ERX_HVDATA 4
BVnumber Binary variable length with maximum length ERX_HVDATA 4
D Date unsigned char [ERX_GET_PACKED_LEN(7)] 3,5,8,12,14,15
F4 Floating point (small) float 11
F8 Floating point (large) double 11
I1 Integer (small) signed char  
I2 Integer (medium) signed short  
I4 Integer (large) signed long  
Knumber Kanji unsigned char [number] 3
KV Kanji variable length ERX_HVDATA 4
KVnumber Kanji variable length with maximum length ERX_HVDATA 3,4
L Logical unsigned char 6
Nnumber1[.number2] Unpacked decimal double 9
unsigned char[number1 + number2] 7,9
NUnumber1[.number2] Unpacked decimal unsigned double 9
unsigned char[number1 + number2] 7,9
Pnumber1[.number2] Packed decimal double 10
unsigned char[ERX_GET_PACKED_LEN(number1 + number2)] 8,10
PUnumber1[.number2] Packed decimal unsigned double 10
unsigned char[ERX_GET_PACKED_LEN(number1 + number2)] 8,10
T Time unsigned char[ERX_GET_PACKED_LEN(13)] 5,8,13,14,16
Unumber Unicode wchar_t[number] 2,17,18
wchar [number + 1] 2,17,18
UV Unicode variable length ERX_HVDATA 4,17
UVnumber Unicode variable length with maximum length ERX_HVDATA 4,17,19

See also hints and restrictions on the Software AG IDL data types valid for all programming language bindings under IDL Data Types in the IDL Editor documentation.

Notes:

  1. It is possible to map the data type to the C data type unsigned char[..] without null termination (mainframe Natural, COBOL, PL/I style) or C style string (C data type char[.. + 1] with null termination). The mapping is controlled with the Mapping Options when you generate source files from IDL. See Generate C Source Files from Software AG IDL Files.
  2. It is possible to map the data type to the C data type wchar_t[..] without null termination (mainframe Natural, COBOL, PL/I style) or to a C style wide character string ( C data type wchar_t[.. + 1] with null termination). The mapping is controlled with the Mapping Options when you generate source files from IDL. See Generate C Source Files from Software AG IDL Files.
  3. The field length is given in bytes.
  4. The data type ERX_HVDATA is an RPC-specific data type in the header file erxVData.h for handling variable-length data types. See API Function Descriptions for Variable-length Data Types AV, BV, KV and UV.
  5. Date and Time values are mapped in mainframe-style packed format.
  6. The binary value zero is treated as FALSE. Contents other than a binary zero are treated as TRUE.
  7. Unpacked format is an internal representation of numbers with a specified number of digits used in mainframe environments (Natural, COBOL, PL/I).
  8. Packed format is an internal representation of numbers with a specified number of digits used in mainframe environments (Natural, COBOL, PL/I). ERX_GET_PACKED_LENGTH is a macro within the header file erx.h used to evaluate the length of a field in bytes with packed contents.
  9. It is possible to map the data type to double or unpacked. The mapping is controlled with the Mapping Options when you generate source files from IDL. See Generate C Source Files from Software AG IDL Files.
    • For unpacked, the total number of digits (number1+number2) is 99.

    • For double, it depends on your target environment how many significant digits are supported. In most environments 15 - 16 digits, which means for exact results the total number of digits (number1+number2) should be less than 16.

    If you connect two endpoints, the total number of digits used must be lower or equal than the maxima of both endpoints. For the supported total number of digits for endpoints, see the notes under data types N, NU, P and PU in section Mapping IDL Data Types to target language environment C | CL | COBOL | DCOM | .NET | Java | Natural | PL/I | RPG | XML.

  10. It is possible to map the data type to double or packed. The mapping is controlled with the Mapping Options when you generate source files from IDL. See Generate C Source Files from Software AG IDL Files.
    • For unpacked the total number of digits (number1+number2) is 99.

    • For double, it depends on your target environment how many significant digits are supported. In most environments 15 - 16 digits, which means for exact results the total number of digits (number1+number2) should be less than 16.

    If you connect two endpoints, the total number of digits used must be lower or equal than the maxima of both endpoints. For the supported total number of digits for endpoints, see the notes under data types N, NU, P and PU in section Mapping IDL Data Types to target language environment C | CL | COBOL | DCOM | .NET | Java | Natural | PL/I | RPG | XML.

  11. For float and double, rounding errors can occur, so that the values of senders and receivers might differ.
  12. Count of days AD (anno domini, after the birth of Christ). The valid range is from 1.1.0001 up to 28.11.2737. Mapping of the number to the date in the complete range from 1.1.0001 on, follows the Julian and Gregorian calendar, taking into consideration the following rules:
    1. Years that are evenly divisible by 4 are leap years.
    2. Years that are evenly divisible by 100 are not leap years unless rule 3, below, is true.
    3. Years that are evenly divisible by 400 are leap years.
    4. Before the year 1582 AD, rule 1 from the Julian calendar is used. After the year 1582 AD, rules 1, 2 and 3 of the Gregorian calendar are used.

    See the following table for the relation of the packed number to a real date:

    Date / Range of Dates Value / Range of Values
    1.1.0000 0 (special value - no date)
    undefined dates 1 - 364 (do not use)
    1.1.0001 365
    1.1.1970 719527 (start of C-time functions)
    28.11.2737 999999 (maximum date)
  13. Count of tenth of seconds AD (anno domini, after the birth of Christ). The valid range is from 1.1.0001 00:00:00.0 up to 16.11.3168 9:46:39 plus 0.9 seconds. See the following table for the relation of the packed number to a real time:
    Time / Range of Times Value / Range of Values
    1.1.0000 00:00:00.0 0 (special value - no time)
    undefined times 1 - 315359999
    1.1.0001 00:00:00.0 315360000
    1.1.1970 00:00:00.0 621671328000 (start of C-time functions)
  14. The relation between the packed number of a Date and Time data type is as follows:
    tenths of a second per day =24*60*60*10 = 864000
    number of time = number of date * 864000
    315360000 = 365 * 864000 1.1.0001 00:00:00.0
    621671328000 = 719527 * 864000 1.1.1970 00:00:00.0
    number of date = number of time / 864000  
    365 = 315360000 / 864000 1.1.0001
    719527 = 621671328000 / 864000 1.1.1970
  15. The no date value is the internal state of a #DATE (Natural type D) variable after a RESET #DATE is executed within Natural programs. This internal state is not a valid date. The no date value can be transferred as the invalid date 1.1.0 from RPC clients to servers and vice versa. C Wrapper supports Natural no date value. C Wrapper passes 0 to the C application when no date is received. With the same value of 0, the C application can send no date to its partner (client or server).
  16. The no time value is the internal state of a #TIME (Natural type T) variable after a RESET #TIME is executed within Natural programs. This internal state is not a valid time. The no time value can be transferred as the invalid time 1.1.0 0:00:00.0 from RPC clients to servers and vice versa. C Wrapper supports Natural no time value. C Wrapper passes 0 to the C application when no time is received. With the same value of 0, the C application can send no time to its partner (client or server).
  17. The Unicode encoding provided (on receive) or expected (on send) on the API depends on the width of the wchar_t data type in your environment:
    Size of wchar_t UTF Version Used
    2 bytes (Windows and some UNIX environments) UTF-16
    4 bytes (some other UNIX environments) UTF-32

    The endianness (big endian or little endian) of the Unicode encoding that is used (UTF-16-LE, UTF-16-BE, UTF32-LE or UTF32-BE) is the same as the hardware architecture of the machine.

  18. In environments where sizeof(wchar_t) is 4 bytes, only Unicode characters from the Basic Multilingual Plane are supported (code points U+0000 to U+FFFF except U+D800 to U+DFFF reserved for leading and trailing surrogates). Unicode characters from Supplementary Planes (code points U+10000 to U+10FFFF) are not supported.
  19. In environments where sizeof(wchar_t) is 4 bytes, the number of UTF-16 Unicode code points (after character conversion from UTF-32 to UTF-16) must be less than or equal to number. Note the following:
    • Unicode characters from the Basic Multilingual Plane (code points U+0000 to U+FFFF except U+D800 to U+DFFF, reserved for leading and trailing surrogates) require one code point in UTF-16

    • Unicode characters from Supplementary Planes (code points U+10000 to U+10FFFF) require two code points in UTF-16

Mapping Library Name and Alias

The library name as specified in the IDL file is sent from a client to the server. Special characters are not replaced. The library alias is not sent to the server.

In the RPC server, the IDL library name sent may be used to locate the target server. See Locating and Calling the Target Server under z/OS (CICS, Batch, IMS) | C | .NET | BS2000.

The library name as given in the IDL file is used to compose the names of the generated output files. See library-definition under Software AG IDL Grammar in the IDL Editor documentation. Therefore the allowed characters are restricted by the underlying file system.

For the server interface object, the name is composed with a prefix D as Dlibrary-name.c and for the server as library-name.c. For the client interface object the name is composed with a prefix C as Clibrary name.c. For both interface objects the same header file Clibrary-name.h is also used. When the name of the generated sources is built, lower and uppercase characters are considered and the special characters '#', '$', '&', '+', '-', '.', '/' and '@' used in the name for libraries are replaced by the character underscore '_'. Other special characters used in the library name are not changed and may lead to problems with your underlying file system.

Aliases for the library name in the IDL file are not supported in C Wrapper language binding. See library-definition.

Examples:

  • A library name of #HU$GO. results in C_HU_GO_.c and C_HU_GO_.h as the client interface object file name for the generated source.

  • A library name of #HU$GO. results in D_HU_GO_.c and C_HU_GO_.h as the server interface object file name for the generated source.

Mapping Program Name and Alias

The program name is sent from a client to the server. Special characters are not replaced. The program alias is not sent to the server.

In the RPC server, the IDL program name sent is used to locate the target server. See Locating and Calling the Target Server under z/OS (CICS, Batch, IMS) | C | .NET | BS2000.

The program names as given in the IDL file are mapped to functions within the generated C sources. See program-definition under Software AG IDL Grammar in the IDL Editor documentation. When building function names, lower and uppercase characters are considered and the special characters '#', '$', '&', '+', '-', '.', '/' and '@' are replaced by the character underscore '_' valid for C names. Other special characters used in the program name are not changed and may lead to compilation errors when compiling the generated sources.

Aliases for the program name in the program-definition are not supported in C Wrapper language binding.

Example

  • A parameter name of #HU$GO. results in _HU_GO_ as the function name for the C programming language.

Mapping Parameter Names

The parameter names as given in the IDL file are mapped to parameters of the generated C functions. See parameter-data-definition under Software AG IDL Grammar in the IDL Editor documentation. When building parameters, lower and uppercase characters are considered and the special characters '#', '$', '&', '+', '-', '.', '/' and '@' are replaced by the character underscore '_' valid for C names.

  • Examples:
    A parameter name of #HU$GO. results in _HU_GO_ as the parameter name for the C programming language.

Mapping Fixed and Unbounded Arrays

  • Fixed arrays within the IDL file are mapped to fixed C arrays. The upper bound given in the IDL file is decremented by 1 because C arrays always start with the lower bound 0. For example the number (I2/5) in the IDL file will be mapped to signed short number [4].

    See the array-definition under Software AG IDL Grammar in the IDL Editor documentation for the syntax on how to describe fixed arrays within the IDL file and refer to fixed-bound-array-index.

  • Unbounded arrays within the IDL file are mapped to the ERX_HARRAY data type found in the header file erxArray.h. See Using Unbounded Arrays for more information.

    See the array-definition for the syntax of unbounded arrays within the IDL file and refer to unbounded-array-index.

Mapping Groups and Periodic Groups

Groups within the IDL file are mapped to the C data type struct. See the group-parameter-definition under Software AG IDL Grammar in the IDL Editor documentation for the syntax on how to describe groups within the IDL file.

Mapping Structures

Structures within the IDL file are mapped to the C data type struct like groups. See structure-definition for the syntax on how to describe structures within the IDL file.

Mapping the Direction Attributes In, Out, InOut

The IDL syntax allows you to define parameters as IN parameters, OUT parameters, or IN OUT parameters (which is the default if nothing is specified). This direction specification is reflected in the generated C interface object as follows:

  • Parameters with the IN attribute are sent from the RPC client to the RPC server. When the parameter is a simple parameter (that is, no fixed or unbounded array, no group and no structure) the parameter is provided with the method call by value. Complex parameters such as fixed and unbounded arrays, groups and structures are provided with the call by reference method.

  • Parameters with the OUT attribute are sent from the RPC server to the RPC client. They are always provided with the call by reference method.

  • Parameters with the IN and OUT attribute are sent from the RPC client to the RPC server and then back to the RPC client. They are always provided with the call by reference method.

Note that only the direction information of the top-level fields (level 1) is relevant. Group fields always inherit the specification from their parent. A different specification is ignored.

See the attribute-list for the syntax on how to describe attributes within the IDL file and refer to direction attribute.

Mapping the ALIGNED Attribute

The ALIGNED Attribute is not relevant for the programming language C. However, a C client can send the ALIGNED attribute to an RPC server where it might be needed.

See the attribute-list for the syntax of attributes in the IDL file and refer to the aligned attribute.

Calling Servers as Procedures or Functions

The IDL syntax allows definitions of procedures only. It does not have the concept of a function. A function is a procedure which, in addition to the parameters, returns a value. Procedures and functions are transparent between clients and servers. This means a client using a function can call a server implemented as a procedure and vice versa.

It is possible to call the remote procedure as a function and not as a procedure, if you prefer it and if it suits your interface.

Example

The function float sin(float x) will be called as a function - and not as a procedure - when defined in the IDL file as follows:

Library ... is 
  Program 'sin' is 
    Define Data Parameter 
      1 x (F4) In 
      1 Function_Result (F4) Out 
    End-Define

it can be invoked as :

y = sin(x);

When you generate source files from IDL (see Generate C Source Files from Software AG IDL Files), a C function instead of a C procedure is generated if the following is true for the interface description in the IDL file:

  • The last parameter's name is function_result. The name function_result is case-insensitive.

  • The last parameter's direction is Out. See attribute-list.

  • The last parameter is a scalar variable, that is, not an array.

  • The last parameter is of type and length:

    Software AG IDL Description Note
    I1 Integer (small) 2
    I2 Integer (medium) 2
    I4 Integer (large) 2
    A1 Alphanumeric with length 1 2
    B1 Binary with length 1 2
    L Logical 2
    Nnumber [.number] Unpacked decimal 1, 2
    Pnumber [.number] Packed decimal 1, 2
    NUnumber [.number] Unpacked decimal unsigned 1, 2
    PUnumber [.number] Packed decimal unsigned 1, 2

    Notes:

    1. The data types must be mapped to double. See Mapping Options when you generate C source files from Software AG IDL files.
    2. The type of the function returned is defined by Mapping IDL Data Types to C Data Types.