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
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
|
SWI-Prolog and the GMP library
Author: Jan Wielemaker
Created: Aug 21, 2005
Udated: Nov 12, 2011, document creating gmp.lib
The GNU GMP library (GNU Multiple Precision Arithmetic Library), is
available from http://gmplib.org/. It provides multiple precision
integer and rational (N/M) arithmetic to SWI-Prolog. By default,
SWI-Prolog uses GMP, but it can be built without using the configure
option --disable-gmp. Statically linking GMP makes SWI-Prolog about 20%
larger, which may be the most important reason not to want it.
Unix
----
On Unix(-like) systems, configure should deal with building and
installation issues.
Windows
-------
Windows is a different matter. The GMP people state it is too much
trouble making a Windows version. There are several ways to get the
official sources compiled, notably using MinGW [1]. An easier
alternative is to use the fork created by the MPIR project, which can be
found at http://mpir.org/.
That said, MPIR requires VS, which requires a lot of other stuff to
install, while the whole lot comes nicely installed with MinGW and its
cross-compilers. The current gmp.lib is produced on an Ubuntu 11.04
system using these commands and the GMP 5.0.2 sources:
% ./configure --host=i586-mingw32msvc --prefix=$HOME/mingw \
--disable-shared --enable-static
% make
% make install
% cd ~/mingw/lib
% GCC=i586-mingw32msvc-gcc
% NM=i586-mingw32msvc-nm
% AR=i586-mingw32msvc-ar
% RANLIB=i586-mingw32msvc-ranlib
% $AR x `$GCC print-libgcc-file-name` _chkstk.o
% $CC -c ms_snprintf.c
% $AR q gmp.a _chkstk.o ms_snprintf.o
% $RANLIB libgmp.a
% mv gmp.a gmp.lib
Note that _chkstk.o is the archive member holding alloca(), which can be
verified using
% $NM -o `$GCC -print-libgcc-file-name` | grep alloca
ms_snprintf.c is here, including the motivation why it is needed.
================
/* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
MS-Windows _snprintf() may look like C99 snprintf(), but is is not quite
the same: on overflow, the buffer is *not* 0-terminated and the return
is negative (unspecified how negative). The code below works around
this, returning count on overflow. This is still not the same as the C99
version that returns the number of characters that would have been
written, but it seems to be enough for our purposes.
See http://www.di-mgt.com.au/cprog.html#snprintf
The above came from the provided link, but it is even worse (copied from
VS2005 docs):
- If len < count, then len characters are stored in buffer, a
null-terminator is appended, and len is returned.
- If len = count, then len characters are stored in buffer, no
null-terminator is appended, and len is returned.
- If len > count, then count characters are stored in buffer, no
null-terminator is appended, and a negative value is returned.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */
#include <stdarg.h>
#include <stdlib.h>
int
snprintf(char *buffer, size_t count, const char *fmt, ...)
{ va_list ap;
int ret;
va_start(ap, fmt);
ret = _vsnprintf(buffer, count-1, fmt, ap);
va_end(ap);
if ( ret < 0 || ret == count )
{ ret = count;
buffer[count-1] = '\0';
}
return ret;
}
================
=====
[1]. This readme used to document these. Most likely this documentation
is out of date, but you may want to check versions ofthe file pior
to Sep 22, 2011.
|
