The header
<tgmath.h> provides type-generic macros for
<math.h> and
<complex.h> functions that have
float (suffixed with
f),
double, and
long double (suffixed with
l) versions. The arguments that vary across the three functions and have type
float,
double, and
long double, respectively, are called
generic arguments.
The following rules describe which function is actually called if a type-generic macro is invoked. If any generic argument has type
long double or
long double complex, the
long double function is called. Else, if any generic argument has type
double,
double complex, or an integer type, the
double version is invoked. Otherwise, the macro expands to the
float implementation.
For the macros in the following table, both real and complex functions exist. The real functions are prototyped in
<math.h> and the complex equivalents in
<complex.h>. The complex function is called if any of the generic arguments is a complex value. Otherwise, the real equivalent is called.
MACRO
REAL FUNCTION
COMPLEX FUNCTION
No complex functions exist for the following macros, so passing a complex value to a generic argument invokes undefined behaviour:
atan2()
fma()
llround()
remainder()
cbrt()
fmax()
log10()
remquo()
ceil()
fmin()
log1p()
rint()
copysign()
fmod()
log2()
round()
erf()
frexp()
logb()
scalbn()
erfc()
hypot()
lrint()
scalbln()
exp2()
ilogb()
lround()
tgamma()
expm1()
ldexp()
nextbyint()
trunc()
fdim()
lgamma()
nextafter()
floor()
llrint()
nexttoward()
The following macros always expand to a complex function:
carg()
cimag()
conj()
cproj()
creal()
This header includes
<complex.h> and
<math.h>.