|
ActiveTcl User Guide |
|
Tcl_CreateMathFunc, Tcl_GetMathFuncInfo, Tcl_ListMathFuncs -
Define, query and enumerate math functions for expressions
#include <tcl.h>
void
Tcl_CreateMathFunc(interp, name, numArgs, argTypes, proc,
clientData)
int
Tcl_GetMathFuncInfo(interp, name, numArgsPtr,
argTypesPtr, procPtr, clientDataPtr)
Tcl_Obj *
Tcl_ListMathFuncs(interp, pattern)
- Tcl_Interp *interp
(in)
- Interpreter in which new function will be defined.
- CONST char *name (in)
- Name for new function.
- int numArgs (in)
- Number of arguments to new function; also gives size of
argTypes array.
- Tcl_ValueType *argTypes (in)
- Points to an array giving the permissible types for each
argument to function.
- Tcl_MathProc *proc (in)
- Procedure that implements the function.
- ClientData clientData (in)
- Arbitrary one-word value to pass to proc when it is
invoked.
- int *numArgsPtr (out)
- Points to a variable that will be set to contain the number of
arguments to the function.
- Tcl_ValueType **argTypesPtr (out)
- Points to a variable that will be set to contain a pointer to
an array giving the permissible types for each argument to the
function which will need to be freed up using Tcl_Free.
- Tcl_MathProc **procPtr (out)
- Points to a variable that will be set to contain a pointer to
the implementation code for the function (or NULL if the function
is implemented directly in bytecode.)
- ClientData *clientDataPtr (out)
- Points to a variable that will be set to contain the clientData
argument passed to Tcl_CreateMathFunc when the function was
created if the function is not implemented directly in
bytecode.
- CONST char *pattern (in)
- Pattern to match against function names so as to filter them
(by passing to Tcl_StringMatch), or NULL to not
apply any filter.
Tcl allows a number of mathematical functions to be used in
expressions, such as sin, cos, and hypot.
Tcl_CreateMathFunc allows applications to add additional
functions to those already provided by Tcl or to replace existing
functions. Name is the name of the function as it will
appear in expressions. If name doesn't already exist as a
function then a new function is created. If it does exist, then the
existing function is replaced. NumArgs and argTypes
describe the arguments to the function. Each entry in the
argTypes array must be one of TCL_INT, TCL_DOUBLE,
TCL_WIDE_INT, or TCL_EITHER to indicate whether the corresponding
argument must be an integer, a double-precision floating value, a
wide (64-bit) integer, or any, respectively.
Whenever the function is invoked in an expression Tcl will
invoke proc. Proc should have arguments and result
that match the type Tcl_MathProc:
typedef int Tcl_MathProc(
ClientData clientData,
Tcl_Interp *interp,
Tcl_Value *args,
Tcl_Value *resultPtr);
When proc is invoked the clientData and
interp arguments will be the same as those passed to
Tcl_CreateMathFunc. Args will point to an array of
numArgs Tcl_Value structures, which describe the actual
arguments to the function:
typedef struct Tcl_Value {
Tcl_ValueType type;
long intValue;
double doubleValue;
Tcl_WideInt wideValue;
} Tcl_Value;
The type field indicates the type of the argument and is
one of TCL_INT, TCL_DOUBLE or TCL_WIDE_INT. It will match the
argTypes value specified for the function unless the
argTypes value was TCL_EITHER. Tcl converts the argument
supplied in the expression to the type requested in
argTypes, if that is necessary. Depending on the value of
the type field, the intValue, doubleValue or
wideValue field will contain the actual value of the
argument.
Proc should compute its result and store it either as an
integer in resultPtr->intValue or as a floating value in
resultPtr->doubleValue. It should set also
resultPtr->type to one of TCL_INT, TCL_DOUBLE or
TCL_WIDE_INT to indicate which value was set. Under normal
circumstances proc should return TCL_OK. If an error occurs
while executing the function, proc should return TCL_ERROR
and leave an error message in the interpreter's result.
Tcl_GetMathFuncInfo retrieves the values associated with
function name that were passed to a preceding
Tcl_CreateMathFunc call. Normally, the return code is
TCL_OK but if the named function does not exist,
TCL_ERROR is returned and an error message is placed in the
interpreter's result.
If an error did not occur, the array reference placed in the
variable pointed to by argTypesPtr is newly allocated, and
should be released by passing it to Tcl_Free. Some functions (the
standard set implemented in the core) are implemented directly at
the bytecode level; attempting to retrieve values for them causes a
NULL to be stored in the variable pointed to by procPtr and
the variable pointed to by clientDataPtr will not be
modified.
Tcl_ListMathFuncs returns a Tcl object containing a list
of all the math functions defined in the interpreter whose name
matches pattern. In the case of an error, NULL is returned
and an error message is left in the interpreter result, and
otherwise the returned object will have a reference count of
zero.
expression, mathematical function
expr, info, Tcl_Free, Tcl_NewListObj
Copyright © 1995-1997 Roger E. Critchlow Jr.
Copyright © 1989-1993 The Regents of the University of
California. Copyright © 1994-1996 Sun Microsystems, Inc.