C HDS Structures

This Appendix describes the HDS structures used by the TRANSFORM facility.⁸ There are three of these, distinguished by their HDS type:

TYPE = TRN_TRANSFORM

structures are used to hold data defining a complete (possibly multi-stage) transformation and are handled directly by user-level TRANSFORM routines.

TYPE = TRN_MODULE

structures are used to hold transformation module information which represents a simple 1-stage transformation without associated classification information. These structures are not handled directly by user-level TRANSFORM routines.

TYPE = TRN_CLASS

structures are used to hold classification information about the properties of a transformation. These structures are also not handled directly by the user-level TRANSFORM routines.

The following Sections describe these structures in detail.

C.1 The TRN_TRANSFORM Structure

An HDS structure of type TRN_TRANSFORM contains the complete specification of a (possibly multi-stage) transformation with its associated classification information. It is defined as follows:

The structure components have the following meanings and restrictions:

TRN_VERSION

is a mandatory scalar _REAL component containing the version number of the TRANSFORM software which created or last modified the TRN_TRANSFORM structure or any of its components. This information allows programs linked with old versions of TRANSFORM software to detect if they are processing data structures created by more recent versions, with which they may not be compatible.

FORWARD

is a mandatory scalar _CHAR component which must contain one of the two values ‘DEFINED’ or ‘UNDEFINED’ (case insensitive). It specifies whether the transformation’s forward mapping is defined or undefined.

INVERSE

is a mandatory scalar _CHAR component which must contain one of the two values ‘DEFINED’ or ‘UNDEFINED’ (case insensitive). It specifies whether the transformation’s inverse mapping is defined or undefined. It is not permitted for both the FORWARD and INVERSE components to have the value ‘UNDEFINED’.

MODULE_ARRAY()

is a mandatory 1-dimensional array of structures of type TRN_MODULE. It holds a sequence of transformation modules which are to be used in succession to define the overall transformation. If the FORWARD component has the value ‘DEFINED’, then all the modules in this array must have their forward mappings defined. Similarly, if the INVERSE component has the value ‘DEFINED’, then all the modules must have their inverse mappings defined. Adjacent modules in this array must have matching numbers of output and input variables. The number of input variables for the transformation as a whole is determined by the number of input variables for the first module in this array, while the number of output variables is determined by the number of output variables for the final module.

CLASSIFICATION

is an optional scalar structure component of type TRN_CLASS. If present, it holds the classification information for the entire transformation. If absent, a default set of classification properties applies, which corresponds to all elements of the transformation’s logical classification array being set to .FALSE..

C.2 The TRN_MODULE Structure

A structure of type TRN_MODULE contains transformation module information, which represents the simplest form of 1-stage transformation without associated classification information. More complex transformations may be built up by joining several of these modules together, as is done in the MODULE_ARRAY component of a TRN_TRANSFORM structure. The TRN_MODULE structure is defined as follows:

The structure components have the following meanings and restrictions:

NVAR_IN

is a mandatory scalar _INTEGER component with a positive value specifying the number of input variables for the transformation which the module describes.

NVAR_OUT

is a mandatory scalar _INTEGER component with a positive value specifying the number of output variables for the transformation which the module describes.

COMMENT

is an optional scalar _CHAR component of arbitrary length which contains a comment associated with the transformation module. This comment should describe (in English) the type of transformation the module performs. The contents of the COMMENT component are only intended to be human-readable and should not be machine-processed, except that the special three-character sequences ‘-->’ and ‘<--’ (if present) may be automatically interchanged to reflect the effect of inverting the transformation module.

PRECISION

is a mandatory scalar _CHAR component of arbitrary length which contains a valid precision specification for the transformation module, according to the syntax described in Section 5.2.

FORWARD_FUNC()

is a mandatory 1-dimensional _CHAR array, with elements of arbitrary length, containing transformation functions which define the forward mapping (according to the syntax in Appendix A). The number of array elements must match the value stored in the NVAR_OUT component.

INVERSE_FUNC()

is a mandatory 1-dimensional _CHAR array, with elements of arbitrary length, containing transformation functions which define the inverse mapping (according to the syntax in Appendix A). The number of array elements must match the value stored in the NVAR_IN component.

C.3 The TRN_CLASS Structure

A structure of type TRN_CLASS contains information specifying the set of classification properties associated with a transformation. These properties are described in Appendix B. The TRN_CLASS structure is defined as follows:

All the structure components are scalar _LOGICAL objects, whose presence is optional. Their names correspond with the classification properties described in Appendix B and their logical values indicate whether the transformation has the associated property. If any component is absent, its value defaults to .FALSE.