man3/fpclassify.3

   1 .\" Copyright 2002 Walter Harms (walter.harms@informatik.uni-oldenburg.de)
   2 .\" Distributed under GPL, 2002-07-27 Walter Harms
   3 .\" This was done with the help of the glibc manual.
   4 .\"
   5 .\" 2004-10-31, aeb, corrected
   6 .TH FPCLASSIFY 3  2007-07-26 "" "Linux Programmer's Manual"
   7 .SH NAME
   8 fpclassify, isfinite, isnormal, isnan, isinf \- floating-point
   9 classification macros
  10 .SH SYNOPSIS
  11 .nf
  12 .B #include <math.h>
  13 .sp
  14 .BI "int fpclassify(" x );
  15 .sp
  16 .BI "int isfinite(" x );
  17 .sp
  18 .BI "int isnormal(" x );
  19 .sp
  20 .BI "int isnan(" x );
  21 .sp
  22 .BI "int isinf(" x );
  23 .fi
  24 .sp
  25 Link with \fI\-lm\fP.
  26 .sp
  27 .in -4n
  28 Feature Test Macro Requirements for glibc (see
  29 .BR feature_test_macros (7)):
  30 .in
  31 .sp
  32 .\" I haven't fully grokked the source to determine the FTM requirements;
  33 .\" in part, the following has been tested by experiment.
  34 .ad l
  35 .BR fpclassify (),
  36 .BR isfinite (),
  37 .BR isnormal ():
  38 _XOPEN_SOURCE\ >=\ 600 || _ISOC99_SOURCE; or
  39 .I cc\ -std=c99
  40 .br
  41 .BR isnan ():
  42 _BSD_SOURCE || _SVID_SOURCE || _XOPEN_SOURCE || _ISOC99_SOURCE; or
  43 .I cc\ -std=c99
  44 .br
  45 .BR isinf ():
  46 _BSD_SOURCE || _SVID_SOURCE || _XOPEN_SOURCE\ >=\ 600 || _ISOC99_SOURCE; or
  47 .I cc\ -std=c99
  48 .ad b
  49 .SH DESCRIPTION
  50 Floating point numbers can have special values, such as
  51 infinite or NaN.
  52 With the macro
  53 .BI fpclassify( x )
  54 you can find out what type
  55 .I x
  56 is.
  57 The macro takes any floating-point expression as argument.
  58 The result is one of the following values:
  59 .TP
  60 .B FP_NAN
  61 .I x
  62 is "Not a Number".
  63 .TP
  64 .B FP_INFINITE
  65 .I x
  66 is either plus or minus infinity.
  67 .TP
  68 .B FP_ZERO
  69 .I x
  70 is zero.
  71 .TP
  72 .B FP_SUBNORMAL
  73 .I x
  74 is too small to be represented in normalized format.
  75 .TP
  76 .B FP_NORMAL
  77 if nothing of the above is correct then it must be a
  78 normal floating-point number.
  79 .LP
  80 The other macros provide a short answer to some standard questions.
  81 .TP
  82 .BI isfinite( x )
  83 returns a non-zero value if
  84 .br
  85 (fpclassify(x) != FP_NAN && fpclassify(x) != FP_INFINITE)
  86 .TP
  87 .BI isnormal( x )
  88 returns a non-zero value if
  89 (fpclassify(x) == FP_NORMAL)
  90 .TP
  91 .BI isnan( x )
  92 returns a non-zero value if
  93 (fpclassify(x) == FP_NAN)
  94 .TP
  95 .BI isinf( x )
  96 returns 1 if
  97 .I x
  98 is positive infinity, and \-1 if
  99 .I x
 100 is negative infinity.
 101 .SH "CONFORMING TO"
 102 C99
 103 .SH NOTES
 104 In glibc 2.01 and earlier,
 105 .BR isinf ()
 106 returns a non-zero value (actually: 1) if
 107 .I x
 108 is an infinity (positive or negative).
 109 (This is all that C99 requires.)
 110 .SH "SEE ALSO"
 111 .BR finite (3),
 112 .BR INFINITY (3),
 113 .BR isgreater (3)