man3/fpclassify.3

   1 '\" t
   2 .\" Copyright 2002 Walter Harms (walter.harms@informatik.uni-oldenburg.de)
   3 .\"
   4 .\" SPDX-License-Identifier: GPL-1.0-or-later
   5 .\"
   6 .\" This was done with the help of the glibc manual.
   7 .\"
   8 .\" 2004-10-31, aeb, corrected
   9 .TH fpclassify 3 (date) "Linux man-pages (unreleased)"
  10 .SH NAME
  11 fpclassify, isfinite, isnormal, isnan, isinf \- floating-point
  12 classification macros
  13 .SH LIBRARY
  14 Math library
  15 .RI ( libm ", " \-lm )
  16 .SH SYNOPSIS
  17 .nf
  18 .B #include <math.h>
  19 .PP
  20 .BI "int fpclassify(" x );
  21 .BI "int isfinite(" x );
  22 .BI "int isnormal(" x );
  23 .BI "int isnan(" x );
  24 .BI "int isinf(" x );
  25 .fi
  26 .PP
  27 .RS -4
  28 Feature Test Macro Requirements for glibc (see
  29 .BR feature_test_macros (7)):
  30 .RE
  31 .PP
  32 .\" I haven't fully grokked the source to determine the FTM requirements;
  33 .\" in part, the following has been tested by experiment.
  34 .BR fpclassify (),
  35 .BR isfinite (),
  36 .BR isnormal ():
  37 .nf
  38     _ISOC99_SOURCE || _POSIX_C_SOURCE >= 200112L
  39 .fi
  40 .PP
  41 .BR isnan ():
  42 .nf
  43     _ISOC99_SOURCE || _POSIX_C_SOURCE >= 200112L
  44         || _XOPEN_SOURCE
  45         || /* Since glibc 2.19: */ _DEFAULT_SOURCE
  46         || /* glibc <= 2.19: */ _BSD_SOURCE || _SVID_SOURCE
  47 .fi
  48 .PP
  49 .BR isinf ():
  50 .nf
  51     _ISOC99_SOURCE || _POSIX_C_SOURCE >= 200112L
  52         || /* Since glibc 2.19: */ _DEFAULT_SOURCE
  53         || /* glibc <= 2.19: */ _BSD_SOURCE || _SVID_SOURCE
  54 .fi
  55 .SH DESCRIPTION
  56 Floating point numbers can have special values, such as
  57 infinite or NaN.
  58 With the macro
  59 .BI fpclassify( x )
  60 you can find out what type
  61 .I x
  62 is.
  63 The macro takes any floating-point expression as argument.
  64 The result is one of the following values:
  65 .TP 14
  66 .B FP_NAN
  67 .I x
  68 is "Not a Number".
  69 .TP
  70 .B FP_INFINITE
  71 .I x
  72 is either positive infinity or negative infinity.
  73 .TP
  74 .B FP_ZERO
  75 .I x
  76 is zero.
  77 .TP
  78 .B FP_SUBNORMAL
  79 .I x
  80 is too small to be represented in normalized format.
  81 .TP
  82 .B FP_NORMAL
  83 if nothing of the above is correct then it must be a
  84 normal floating-point number.
  85 .PP
  86 The other macros provide a short answer to some standard questions.
  87 .TP 14
  88 .BI isfinite( x )
  89 returns a nonzero value if
  90 .br
  91 (fpclassify(x) != FP_NAN && fpclassify(x) != FP_INFINITE)
  92 .TP
  93 .BI isnormal( x )
  94 returns a nonzero value if
  95 (fpclassify(x) == FP_NORMAL)
  96 .TP
  97 .BI isnan( x )
  98 returns a nonzero value if
  99 (fpclassify(x) == FP_NAN)
 100 .TP
 101 .BI isinf( x )
 102 returns 1 if
 103 .I x
 104 is positive infinity, and \-1 if
 105 .I x
 106 is negative infinity.
 107 .SH ATTRIBUTES
 108 For an explanation of the terms used in this section, see
 109 .BR attributes (7).
 110 .ad l
 111 .nh
 112 .TS
 113 allbox;
 114 lbx lb lb
 115 l l l.
 116 Interface       Attribute       Value
 117 T{
 118 .BR fpclassify (),
 119 .BR isfinite (),
 120 .BR isnormal (),
 121 .BR isnan (),
 122 .BR isinf ()
 123 T}      Thread safety   MT-Safe
 124 .TE
 125 .hy
 126 .ad
 127 .sp 1
 128 .SH STANDARDS
 129 C11, POSIX.1-2008.
 130 .SH HISTORY
 131 POSIX.1-2001, C99.
 132 .PP
 133 In glibc 2.01 and earlier,
 134 .BR isinf ()
 135 returns a nonzero value (actually: 1) if
 136 .I x
 137 is positive infinity or negative infinity.
 138 (This is all that C99 requires.)
 139 .SH NOTES
 140 For
 141 .BR isinf (),
 142 the standards merely say that the return value is nonzero
 143 if and only if the argument has an infinite value.
 144 .SH SEE ALSO
 145 .BR finite (3),
 146 .BR INFINITY (3),
 147 .BR isgreater (3),
 148 .BR signbit (3)