man3/fpclassify.3

   1 .\" Copyright 2002 Walter Harms (walter.harms@informatik.uni-oldenburg.de)
   2 .\"
   3 .\" SPDX-License-Identifier: GPL-1.0-or-later
   4 .\"
   5 .\" This was done with the help of the glibc manual.
   6 .\"
   7 .\" 2004-10-31, aeb, corrected
   8 .TH FPCLASSIFY 3 2021-03-22 "Linux man-pages (unreleased)" "Linux Programmer's Manual"
   9 .SH NAME
  10 fpclassify, isfinite, isnormal, isnan, isinf \- floating-point
  11 classification macros
  12 .SH LIBRARY
  13 Math library
  14 .RI ( libm ", " \-lm )
  15 .SH SYNOPSIS
  16 .nf
  17 .B #include <math.h>
  18 .PP
  19 .BI "int fpclassify(" x );
  20 .BI "int isfinite(" x );
  21 .BI "int isnormal(" x );
  22 .BI "int isnan(" x );
  23 .BI "int isinf(" x );
  24 .fi
  25 .PP
  26 .RS -4
  27 Feature Test Macro Requirements for glibc (see
  28 .BR feature_test_macros (7)):
  29 .RE
  30 .PP
  31 .\" I haven't fully grokked the source to determine the FTM requirements;
  32 .\" in part, the following has been tested by experiment.
  33 .BR fpclassify (),
  34 .BR isfinite (),
  35 .BR isnormal ():
  36 .nf
  37     _ISOC99_SOURCE || _POSIX_C_SOURCE >= 200112L
  38 .fi
  39 .PP
  40 .BR isnan ():
  41 .nf
  42     _ISOC99_SOURCE || _POSIX_C_SOURCE >= 200112L
  43         || _XOPEN_SOURCE
  44         || /* Since glibc 2.19: */ _DEFAULT_SOURCE
  45         || /* Glibc <= 2.19: */ _BSD_SOURCE || _SVID_SOURCE
  46 .fi
  47 .PP
  48 .BR isinf ():
  49 .nf
  50     _ISOC99_SOURCE || _POSIX_C_SOURCE >= 200112L
  51         || /* Since glibc 2.19: */ _DEFAULT_SOURCE
  52         || /* Glibc <= 2.19: */ _BSD_SOURCE || _SVID_SOURCE
  53 .fi
  54 .SH DESCRIPTION
  55 Floating point numbers can have special values, such as
  56 infinite or NaN.
  57 With the macro
  58 .BI fpclassify( x )
  59 you can find out what type
  60 .I x
  61 is.
  62 The macro takes any floating-point expression as argument.
  63 The result is one of the following values:
  64 .TP 14
  65 .B FP_NAN
  66 .I x
  67 is "Not a Number".
  68 .TP
  69 .B FP_INFINITE
  70 .I x
  71 is either positive infinity or negative infinity.
  72 .TP
  73 .B FP_ZERO
  74 .I x
  75 is zero.
  76 .TP
  77 .B FP_SUBNORMAL
  78 .I x
  79 is too small to be represented in normalized format.
  80 .TP
  81 .B FP_NORMAL
  82 if nothing of the above is correct then it must be a
  83 normal floating-point number.
  84 .PP
  85 The other macros provide a short answer to some standard questions.
  86 .TP 14
  87 .BI isfinite( x )
  88 returns a nonzero value if
  89 .br
  90 (fpclassify(x) != FP_NAN && fpclassify(x) != FP_INFINITE)
  91 .TP
  92 .BI isnormal( x )
  93 returns a nonzero value if
  94 (fpclassify(x) == FP_NORMAL)
  95 .TP
  96 .BI isnan( x )
  97 returns a nonzero value if
  98 (fpclassify(x) == FP_NAN)
  99 .TP
 100 .BI isinf( x )
 101 returns 1 if
 102 .I x
 103 is positive infinity, and \-1 if
 104 .I x
 105 is negative infinity.
 106 .SH ATTRIBUTES
 107 For an explanation of the terms used in this section, see
 108 .BR attributes (7).
 109 .ad l
 110 .nh
 111 .TS
 112 allbox;
 113 lbx lb lb
 114 l l l.
 115 Interface       Attribute       Value
 116 T{
 117 .BR fpclassify (),
 118 .BR isfinite (),
 119 .BR isnormal (),
 120 .BR isnan (),
 121 .BR isinf ()
 122 T}      Thread safety   MT-Safe
 123 .TE
 124 .hy
 125 .ad
 126 .sp 1
 127 .SH STANDARDS
 128 POSIX.1-2001, POSIX.1-2008, C99.
 129 .PP
 130 For
 131 .BR isinf (),
 132 the standards merely say that the return value is nonzero
 133 if and only if the argument has an infinite value.
 134 .SH NOTES
 135 In glibc 2.01 and earlier,
 136 .BR isinf ()
 137 returns a nonzero value (actually: 1) if
 138 .I x
 139 is positive infinity or negative infinity.
 140 (This is all that C99 requires.)
 141 .SH SEE ALSO
 142 .BR finite (3),
 143 .BR INFINITY (3),
 144 .BR isgreater (3),
 145 .BR signbit (3)