summaryrefslogtreecommitdiff
path: root/mpfr/BUGS
blob: edb69016d83b4b0e394fe4543bb2d9815d572870 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
Copyright 1999, 2001, 2002, 2003 Free Software Foundation.
Contributed by the Spaces project, INRIA Lorraine.

This file is part of the MPFR Library.

The MPFR Library is free software; you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation; either version 2.1 of the License, or (at your
option) any later version.

The MPFR Library is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public
License for more details.

You should have received a copy of the GNU Lesser General Public License
along with the MPFR Library; see the file COPYING.LIB.  If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA.

##############################################################################

Probably many bugs.

Known bugs:

* The overflows/underflows are not yet implemented in all functions, or
  partially implemented. For instance, mpfr_pow (z, x, y, rnd) fails for
  very small x and some values of y.

* The mpfr_set_ld function assumes that the long double type has an
  exponent of at most 15 bits.

Potential bugs:

* Possible integer overflows on some machines. Some types are used
  inconsistently.

* Possible infinite loop in some functions for particular cases: when
  the exact result is an exactly representable number or the middle of
  consecutive two such numbers.

* mpfr_set_d may give wrong results on some architectures.

* Error analysis for some functions may be incorrect (out-of-date due
  to modifications in the code?).

* mpfr_hypot may fail for x very large, y very small and a very large
  target precision. Other functions may be affected by similar problems.

Problems due to compiler bugs:

* on some architectures (for example alpha-dec-osf), gcc 3.3 wrongly
  compares "long double" floating-point numbers, with optimization level
  1 or higher. This bug can be detected by the following program:

  #include <stdio.h>
  #include <stdlib.h>
  #include <float.h>
  int
  main ()
  {
    long double d;
    d = 1.0; while (d < LDBL_MAX / 2.0) d += d;
    if (d == (long double) 0.0)
      printf ("d equals 0.0\n");
  }

  This results in a problem in the mpfr_set_ld function. A workaround is
  to compile set_ld.c with -O0 (no optimization).