Home | History | Annotate | Line # | Download | only in dist
BUGS revision 1.1.1.1 
      1 Copyright 1999, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011 Free Software Foundation, Inc.
      2 Contributed by the Arenaire and Cacao projects, INRIA.
      3 
      4 This file is part of the GNU MPFR Library.
      5 
      6 The GNU MPFR Library is free software; you can redistribute it and/or modify
      7 it under the terms of the GNU Lesser General Public License as published by
      8 the Free Software Foundation; either version 3 of the License, or (at your
      9 option) any later version.
     10 
     11 The GNU MPFR Library is distributed in the hope that it will be useful, but
     12 WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
     13 or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public
     14 License for more details.
     15 
     16 You should have received a copy of the GNU Lesser General Public License
     17 along with the GNU MPFR Library; see the file COPYING.LESSER.  If not, see
     18 http://www.gnu.org/licenses/ or write to the Free Software Foundation, Inc.,
     19 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA.
     20 
     21 ##############################################################################
     22 
     23 Probably many bugs.
     24 
     25 Known bugs:
     26 
     27 * The overflow/underflow exceptions may be badly handled in some functions;
     28   specially when the intermediary internal results have exponent which
     29   exceeds the hardware limit (2^30 for a 32 bits CPU, and 2^62 for a 64 bits
     30   CPU) or the exact result is close to an overflow/underflow threshold.
     31 
     32 * Under Linux/x86 with the traditional FPU, some functions do not work
     33   if the FPU rounding precision has been changed to single (this is a
     34   bad practice and should be useless, but one never knows what other
     35   software will do).
     36 
     37 * Some functions do not use MPFR_SAVE_EXPO_* macros, thus do not behave
     38   correctly in a reduced exponent range.
     39 
     40 * Function hypot gives incorrect result when on the one hand the difference
     41   between parameters' exponents is near 2*MPFR_EMAX_MAX and on the other hand
     42   the output precision or the precision of the parameter with greatest
     43   absolute value is greater than 2*MPFR_EMAX_MAX-4.
     44 
     45 Potential bugs:
     46 
     47 * Possible incorrect results due to internal underflow, which can lead to
     48   a huge loss of accuracy while the error analysis doesn't take that into
     49   account. If the underflow occurs at the last function call (just before
     50   the MPFR_CAN_ROUND), the result should be correct (or MPFR gets into an
     51   infinite loop). TODO: check the code and the error analysis.
     52 
     53 * Possible integer overflows on some machines.
     54 
     55 * Possible bugs with huge precisions (> 2^30).
     56 
     57 * Possible bugs if the chosen exponent range does not allow to represent
     58   the range [1/16, 16].
     59 
     60 * Possible infinite loop in some functions for particular cases: when
     61   the exact result is an exactly representable number or the middle of
     62   consecutive two such numbers. However for non-algebraic functions, it is
     63   believed that no such case exists, except the well-known cases like cos(0)=1,
     64   exp(0)=1, and so on, and the x^y function when y is an integer or y=1/2^k.
     65 
     66 * The mpfr_set_ld function may be quite slow if the long double type has an
     67   exponent of more than 15 bits.
     68 
     69 * mpfr_set_d may give wrong results on some non-IEEE architectures.
     70 
     71 * Error analysis for some functions may be incorrect (out-of-date due
     72   to modifications in the code?).
     73 
     74 * Possible use of non-portable feature (pre-C99) of the integer division
     75   with negative result.
     76