Type iterator
#include "dip_tpi.h"
Type iterator. For each data type specified by the define DIP_TPI_ALLOW, dip_tpi.h will include the file specified by the define DIP_TPI_FILE. If DIP_TPI_ALLOW is not defined the file will be included for all data types. DIP_TPI_TYPES must be defined as a logical OR of identifier flags and identifier group flags, as given in DIPlib's data types and the table below. During each "iteration" the main symbols defined by dip_tpi.h are DIP_TPI, DIP_TPI_DATA_TYPE, DIP_TPI_IDENTIFIER and DIP_TPI_EXTENSION. The following table shows how these are defined for each data type:
| DIP_TPI | DIP_TPI_DATA_TYPE | DIP_TPI_IDENTIFIER | DIP_TPI_EXTENSION |
| dip_bin8 | DIP_DT_BIN8 | DIP_DTID_BIN8 | _b8 |
| dip_bin16 | DIP_DT_BIN16 | DIP_DTID_BIN16 | _b16 |
| dip_bin32 | DIP_DT_BIN32 | DIP_DTID_BIN32 | _b32 |
| dip_uint8 | DIP_DT_UINT8 | DIP_DTID_UINT8 | _u8 |
| dip_uint16 | DIP_DT_UINT16 | DIP_DTID_UINT16 | _u16 |
| dip_uint32 | DIP_DT_UINT32 | DIP_DTID_UINT32 | _u32 |
| dip_sint8 | DIP_DT_SINT8 | DIP_DTID_SINT8 | _s8 |
| dip_sint16 | DIP_DT_SINT16 | DIP_DTID_SINT16 | _s16 |
| dip_sint32 | DIP_DT_SINT32 | DIP_DTID_SINT32 | _s32 |
| dip_sfloat | DIP_DT_SFLOAT | DIP_DTID_SFLOAT | _sfl |
| dip_dfloat | DIP_DT_DFLOAT | DIP_DTID_DFLOAT | _dfl |
| dip_scomplex | DIP_DT_SCOMPLEX | DIP_DTID_SCOMPLEX | _scx |
| dip_dcomplex | DIP_DT_DCOMPLEX | DIP_DTID_DCOMPLEX | _dcx |
Using this include file it is possible to compile source code for different data types. We recommend that instead of splitting your code into two files, one for generic code and one for type specific code, that you use dip_tpi.h to let the source file include itself. This also prevents dependency problems with makefiles. A source file that includes itself through dip_tpi.h should have the following format:
contents of example.c: #ifndef DIP_TPI #include "diplib.h" #define DIP_TPI_FILE "example.c" #include "dip_tpi.h" /* This is where the generic code should be */ #else /* This is where the type specific code should be */ #endif
In addition to the main defines as described above, there are a number of macro's that are defined by dip_tpi.h:
| DIP_TPI_FUNC ( function name ) | attaches the current type suffix to the function name. |
| DIP_TPI_DEFINE ( function name ) | equivalent to: dip_Error DIP_TPI_FUNC( function name ) useful for function definitions. |
| DIP_TPI_DECLARE ( function name ) | equivalent to: dip_Error DIP_TPI_FUNC( function name ) useful for function declarations. Don't forget the trailing ";". |
| DIP_TPI_NAME ( function name ) | attaches the current type suffix to the function name and puts double quotes around the result, thus creating a string. |
There are also a couple of defines that are only available for some of the data types:
| When DIP_TPI is | |
| dip_sfloat | DIP_TPI_CAST_R2C is defined as dip_scomplex |
| dip_dfloat | DIP_TPI_CAST_R2C is defined as dip_dcomplex |
| dip_scomplex | DIP_TPI_CAST_C2R is defined as dip_sfloat |
| dip_dcomplex | DIP_TPI_CAST_C2R is defined as dip_dfloat |
Other type iterators may be created by making a copy of the dip_tpi.h file and replacing DIP_TPI throughout the file by a different name for the new type iterator.
| Name | Description |
| DIP_TPI_ALLOW | logical OR of data type identifier and identifier group flags to indicate for which data types the file should be included |
| DIP_TPI_FILE | Name of the file to be included by dip_tpi.h |