| [1166] | 1 | <?xml version="1.0" encoding="iso-8859-1"?>
|
|---|
| 2 | <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
|
|---|
| 3 |
|
|---|
| 4 | <!--
|
|---|
| 5 | Copyright 2000-2020 Free Software Foundation, Inc.
|
|---|
| 6 | Contributed by the AriC and Caramba projects, INRIA.
|
|---|
| 7 |
|
|---|
| 8 | This file is part of the GNU MPFR Library.
|
|---|
| 9 |
|
|---|
| 10 | The GNU MPFR Library is free software; you can redistribute it and/or modify
|
|---|
| 11 | it under the terms of the GNU Lesser General Public License as published by
|
|---|
| 12 | the Free Software Foundation; either version 3 of the License, or (at your
|
|---|
| 13 | option) any later version.
|
|---|
| 14 |
|
|---|
| 15 | The GNU MPFR Library is distributed in the hope that it will be useful, but
|
|---|
| 16 | WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
|
|---|
| 17 | or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public
|
|---|
| 18 | License for more details.
|
|---|
| 19 |
|
|---|
| 20 | You should have received a copy of the GNU Lesser General Public License
|
|---|
| 21 | along with the GNU MPFR Library; see the file COPYING.LESSER. If not, see
|
|---|
| 22 | https://www.gnu.org/licenses/ or write to the Free Software Foundation, Inc.,
|
|---|
| 23 | 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA.
|
|---|
| 24 | -->
|
|---|
| 25 |
|
|---|
| 26 | <html xmlns="http://www.w3.org/1999/xhtml" lang="en" xml:lang="en">
|
|---|
| 27 |
|
|---|
| 28 | <head><meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1" />
|
|---|
| 29 | <title>Frequently Asked Questions about GNU MPFR</title>
|
|---|
| 30 | <style type="text/css">/*<![CDATA[*/
|
|---|
| 31 | /* Global stylesheet for visual media */
|
|---|
| 32 |
|
|---|
| 33 | html, body
|
|---|
| 34 | {
|
|---|
| 35 | background: white;
|
|---|
| 36 | color: black;
|
|---|
| 37 | }
|
|---|
| 38 |
|
|---|
| 39 | div.logo { float: right }
|
|---|
| 40 | div.logo img { border: 0 }
|
|---|
| 41 |
|
|---|
| 42 | div.footer img { border: 0 }
|
|---|
| 43 |
|
|---|
| 44 | dt
|
|---|
| 45 | {
|
|---|
| 46 | margin-top: 2ex;
|
|---|
| 47 | margin-bottom: 1ex;
|
|---|
| 48 | font-weight: bolder;
|
|---|
| 49 | }
|
|---|
| 50 |
|
|---|
| 51 | /* For testing: dd { background: #ddddff } */
|
|---|
| 52 |
|
|---|
| 53 | table { margin: 0.5ex auto }
|
|---|
| 54 |
|
|---|
| 55 | li { margin-top: 0.5ex; margin-bottom: 0.5ex }
|
|---|
| 56 |
|
|---|
| 57 | dd + dd
|
|---|
| 58 | {
|
|---|
| 59 | margin-top: 0;
|
|---|
| 60 | margin-bottom: 0;
|
|---|
| 61 | padding-top: 0.5ex;
|
|---|
| 62 | padding-bottom: 0;
|
|---|
| 63 | }
|
|---|
| 64 |
|
|---|
| 65 | li > p, dd > p
|
|---|
| 66 | {
|
|---|
| 67 | margin-top: 0;
|
|---|
| 68 | margin-bottom: 0;
|
|---|
| 69 | padding-top: 0.5ex;
|
|---|
| 70 | padding-bottom: 0.5ex;
|
|---|
| 71 | }
|
|---|
| 72 |
|
|---|
| 73 | .block-code code, code.block-code,
|
|---|
| 74 | .block-code samp, samp.block-code
|
|---|
| 75 | {
|
|---|
| 76 | display: block;
|
|---|
| 77 | padding: 0.5ex 0;
|
|---|
| 78 | margin-left: 2em;
|
|---|
| 79 | }
|
|---|
| 80 |
|
|---|
| 81 | .highlight { background: #dddddd }
|
|---|
| 82 |
|
|---|
| 83 | .nowrap { white-space: nowrap }
|
|---|
| 84 |
|
|---|
| 85 | /*
|
|---|
| 86 | dl.faq { counter-reset: faq }
|
|---|
| 87 |
|
|---|
| 88 | dl.faq dt:before
|
|---|
| 89 | {
|
|---|
| 90 | counter-increment: faq;
|
|---|
| 91 | content: counter(faq) ". ";
|
|---|
| 92 | }
|
|---|
| 93 | */
|
|---|
| 94 |
|
|---|
| 95 | dl.faq dt { background: #dddddd }
|
|---|
| 96 |
|
|---|
| 97 | dl.faq dd
|
|---|
| 98 | {
|
|---|
| 99 | border-left: 4px solid;
|
|---|
| 100 | border-color: transparent;
|
|---|
| 101 | margin-left: 0em;
|
|---|
| 102 | padding-left: 1.5em;
|
|---|
| 103 | }
|
|---|
| 104 |
|
|---|
| 105 | dl.faq dt:target + dd
|
|---|
| 106 | {
|
|---|
| 107 | border-left-color: #aaaaaa;
|
|---|
| 108 | }
|
|---|
| 109 |
|
|---|
| 110 | var.env { font-style: normal }
|
|---|
| 111 | /*]]>*/</style>
|
|---|
| 112 | </head>
|
|---|
| 113 |
|
|---|
| 114 | <body>
|
|---|
| 115 |
|
|---|
| 116 | <h1>Frequently Asked Questions about <cite><acronym>GNU</acronym> <acronym>MPFR</acronym></cite></h1>
|
|---|
| 117 |
|
|---|
| 118 | <p><strong>Important notice: Problems with a particular version of
|
|---|
| 119 | <cite><acronym>MPFR</acronym></cite> are discussed in the corresponding
|
|---|
| 120 | bugs page.</strong></p>
|
|---|
| 121 |
|
|---|
| 122 | <p>The latest version of this <acronym>FAQ</acronym> is available at
|
|---|
| 123 | <a href="https://www.mpfr.org/faq.html">https://www.mpfr.org/faq.html</a>.
|
|---|
| 124 | Please look at this version if possible.</p>
|
|---|
| 125 |
|
|---|
| 126 | <ol>
|
|---|
| 127 | <li><a href="#mpfr_vs_mpf">What are the differences between
|
|---|
| 128 | <cite><acronym>MPF</acronym></cite> from <cite><acronym>GMP</acronym></cite>
|
|---|
| 129 | and <cite><acronym>MPFR</acronym></cite>?</a></li>
|
|---|
| 130 | <li><a href="#mpf2mpfr">How to convert my program written using
|
|---|
| 131 | <cite><acronym>MPF</acronym></cite> to
|
|---|
| 132 | <cite><acronym>MPFR</acronym></cite>?</a></li>
|
|---|
| 133 | <li><a href="#no_libgmp">At configure time, I get the error: <q>libgmp not found or uses a different ABI.</q></a></li>
|
|---|
| 134 | <li><a href="#undef_ref1">I get undefined reference to <code>__gmp_get_memory_functions</code>.</a></li>
|
|---|
| 135 | <li><a href="#undef_ref2">When I link my program with
|
|---|
| 136 | <cite><acronym>MPFR</acronym></cite>, I get undefined reference
|
|---|
| 137 | to <code>__gmpXXXX</code>.</a></li>
|
|---|
| 138 | <li><a href="#crash_high_prec">My program crashes with high precisions.</a></li>
|
|---|
| 139 | <li><a href="#accuracy">Though I have increased the precision, the results
|
|---|
| 140 | are not more accurate.</a></li>
|
|---|
| 141 | <li><a href="#detect_mpfr">How can I detect <cite><acronym>MPFR</acronym></cite>
|
|---|
| 142 | installation using <cite>autoconf</cite> or <cite>pkg-config</cite>?</a></li>
|
|---|
| 143 | <li><a href="#cite">How to cite <cite><acronym>MPFR</acronym></cite> in a
|
|---|
| 144 | scientific publication?</a></li>
|
|---|
| 145 | <li><a href="#fpic">When I build <cite><acronym>MPFR</acronym></cite>, I get
|
|---|
| 146 | an error asking me to recompile with <samp>-fPIC</samp>.</a></li>
|
|---|
| 147 | </ol>
|
|---|
| 148 |
|
|---|
| 149 | <dl class="faq">
|
|---|
| 150 |
|
|---|
| 151 | <dt id="mpfr_vs_mpf">1. What are the differences between
|
|---|
| 152 | <cite><acronym>MPF</acronym></cite> from <cite><acronym>GMP</acronym></cite>
|
|---|
| 153 | and <cite><acronym>MPFR</acronym></cite>?</dt>
|
|---|
| 154 |
|
|---|
| 155 | <dd><p>The main differences are:</p>
|
|---|
| 156 | <ul>
|
|---|
| 157 | <li><p>The precision of a <cite><acronym>MPFR</acronym></cite> variable
|
|---|
| 158 | is the <em>exact</em> number of bits used for its mantissa, whereas in
|
|---|
| 159 | <cite><acronym>MPF</acronym></cite>, the precision requested by the user
|
|---|
| 160 | is a minimum value (<cite><acronym>MPF</acronym></cite> generally uses a
|
|---|
| 161 | higher precision). With the additional difference below, this implies that
|
|---|
| 162 | the <cite><acronym>MPFR</acronym></cite> results do not depend on the
|
|---|
| 163 | number of bits (16, 32, 64 or more) of the underlying architecture.</p></li>
|
|---|
| 164 | <li><p>As a consequence, <cite><acronym>MPFR</acronym></cite> uses a
|
|---|
| 165 | base-2 exponent, whereas in <cite><acronym>MPF</acronym></cite>, this
|
|---|
| 166 | is a base-2<sup>32</sup> or base-2<sup>64</sup> exponent, depending on
|
|---|
| 167 | the limb size. For this reason (and other internal ones), the maximum
|
|---|
| 168 | exponent range in <cite><acronym>MPFR</acronym></cite> is different
|
|---|
| 169 | (and smaller, if the exponent is represented by the same type as in
|
|---|
| 170 | <cite><acronym>MPF</acronym></cite>).</p></li>
|
|---|
| 171 | <li><p><cite><acronym>MPFR</acronym></cite> provides an additional rounding
|
|---|
| 172 | mode argument to its functions; furthermore, it is guaranteed that the
|
|---|
| 173 | result of any operation is the nearest possible floating-point value from
|
|---|
| 174 | the exact result (considering the input variables as exact values), taking
|
|---|
| 175 | into account the precision of the destination variable and the rounding
|
|---|
| 176 | mode. <cite><acronym>MPFR</acronym></cite> also says whether the rounded
|
|---|
| 177 | result is above or below the exact result.</p></li>
|
|---|
| 178 | <li><p><cite><acronym>MPFR</acronym></cite> supports much more functions
|
|---|
| 179 | (in particular transcendental functions such as exponentials, logarithms,
|
|---|
| 180 | trigonometric functions and so on) and special values: signed zeros,
|
|---|
| 181 | infinities, not-a-number (NaN).</p></li>
|
|---|
| 182 | </ul></dd>
|
|---|
| 183 |
|
|---|
| 184 | <dt id="mpf2mpfr">2. How to convert my program written using
|
|---|
| 185 | <cite><acronym>MPF</acronym></cite> to
|
|---|
| 186 | <cite><acronym>MPFR</acronym></cite>?</dt>
|
|---|
| 187 |
|
|---|
| 188 | <dd><p>You need to add <q><code>r</code></q> to the function names, and to
|
|---|
| 189 | specify the rounding mode (<code>MPFR_RNDN</code> for rounding to nearest,
|
|---|
| 190 | <code>MPFR_RNDZ</code> for rounding toward zero, <code>MPFR_RNDU</code>
|
|---|
| 191 | for rounding toward plus infinity, <code>MPFR_RNDD</code> for rounding
|
|---|
| 192 | toward minus infinity). You can also define macros as follows:
|
|---|
| 193 | <code class="block-code">#define mpf_add(a, b, c) mpfr_add(a, b, c, MPFR_RNDN)</code></p>
|
|---|
| 194 | <p>The header file <samp>mpf2mpfr.h</samp> from the
|
|---|
| 195 | <cite><acronym>MPFR</acronym></cite> distribution automatically
|
|---|
| 196 | redefines all <cite><acronym>MPF</acronym></cite> functions in this
|
|---|
| 197 | way, using the default <cite><acronym>MPFR</acronym></cite> rounding
|
|---|
| 198 | mode. Thus you simply need to add the following line in all your files
|
|---|
| 199 | using <cite><acronym>MPF</acronym></cite> functions:
|
|---|
| 200 | <code class="block-code">#include <mpf2mpfr.h></code>
|
|---|
| 201 | just after the <samp>gmp.h</samp> and <samp>mpfr.h</samp>
|
|---|
| 202 | header files. If the program uses <cite><acronym>MPF</acronym></cite>
|
|---|
| 203 | internals (such as direct access to <code>__mpf_struct</code> members),
|
|---|
| 204 | additional changes will be needed.</p></dd>
|
|---|
| 205 |
|
|---|
| 206 | <dt id="no_libgmp">3. At configure time, I get the error: <q>libgmp not found or uses a different ABI.</q></dt>
|
|---|
| 207 |
|
|---|
| 208 | <dd><p>This test (<samp>checking for __gmpz_init in -lgmp</samp>) comes
|
|---|
| 209 | after the <samp>gmp.h</samp> detection. The failure occurs either because
|
|---|
| 210 | the <cite><acronym>GMP</acronym></cite> library could not be found
|
|---|
| 211 | (as it is not in the provided library search paths) or because the
|
|---|
| 212 | <cite><acronym>GMP</acronym></cite> library that was found does not have
|
|---|
| 213 | the expected <acronym title="Application Binary Interface">ABI</acronym>
|
|---|
| 214 | (<abbr>e.g.</abbr> 32-bit <abbr>vs</abbr> 64-bit). The former problem can be
|
|---|
| 215 | due to the fact that a static build of <cite><acronym>MPFR</acronym></cite>
|
|---|
| 216 | was requested while only a shared <cite><acronym>GMP</acronym></cite> library
|
|---|
| 217 | is installed (or the opposite, but another error can also show up in this
|
|---|
| 218 | case, see the <a href="#fpic">question about <samp>-fPIC</samp></a>). The
|
|---|
| 219 | latter problem can have several causes:</p>
|
|---|
| 220 | <ul>
|
|---|
| 221 | <li>A wrong libgmp library has been picked up. This can occur if you have
|
|---|
| 222 | several <cite><acronym>GMP</acronym></cite> versions installed on the
|
|---|
| 223 | machine and something is wrong with the provided library search paths.</li>
|
|---|
| 224 | <li>Wrong compiler options (<samp>CFLAGS</samp>) were given. In general, the
|
|---|
| 225 | presence or absence of the <samp>-m64</samp> compiler option must match the
|
|---|
| 226 | library <acronym title="Application Binary Interface">ABI</acronym>.</li>
|
|---|
| 227 | <li>A wrong <samp>gmp.h</samp> file has been picked up (if you have several
|
|---|
| 228 | <cite><acronym>GMP</acronym></cite> versions installed). Indeed, by default,
|
|---|
| 229 | <cite><acronym>MPFR</acronym></cite> gets the compiler options from the
|
|---|
| 230 | <samp>gmp.h</samp> file (with <cite><acronym>GMP</acronym></cite> 4.2.3
|
|---|
| 231 | or later); this is needed because <cite><acronym>GMP</acronym></cite> does
|
|---|
| 232 | not necessarily use the default <acronym>ABI</acronym>. The consequence is
|
|---|
| 233 | that if the <samp>gmp.h</samp> file is associated with a library using a
|
|---|
| 234 | different <acronym>ABI</acronym>, the <acronym>ABI</acronym>-related options
|
|---|
| 235 | will be incorrect. Hence the failure.</li>
|
|---|
| 236 | </ul>
|
|---|
| 237 | <p>Note: The <samp>config.log</samp> output gives more information
|
|---|
| 238 | than the error message. In particular, see the output of the test:
|
|---|
| 239 | <samp>checking for CC and CFLAGS in gmp.h</samp>; it should give you
|
|---|
| 240 | the default compiler options (from <samp>gmp.h</samp>).</p>
|
|---|
| 241 |
|
|---|
| 242 | <p>See also the answer to the <a href="#undef_ref1">next question</a>.</p></dd>
|
|---|
| 243 |
|
|---|
| 244 | <dt id="undef_ref1">4. I get undefined reference to <code>__gmp_get_memory_functions</code>.</dt>
|
|---|
| 245 |
|
|---|
| 246 | <dd><p>Note: this was mainly a problem when upgrading from
|
|---|
| 247 | <cite><acronym>GMP</acronym></cite> 4.1.4 to a later version,
|
|---|
| 248 | but information given below may still be useful in other cases,
|
|---|
| 249 | when several <cite><acronym>GMP</acronym></cite> libraries are
|
|---|
| 250 | installed on the same machine.</p>
|
|---|
| 251 |
|
|---|
| 252 | <p>If you get such an error, in particular when running
|
|---|
| 253 | <samp>make check</samp>, then this probably means that you are using
|
|---|
| 254 | the header file from <cite><acronym>GMP</acronym></cite> 4.2.x but the
|
|---|
| 255 | <cite><acronym>GMP</acronym></cite> 4.1.4 library. This can happen if
|
|---|
| 256 | several <cite><acronym>GMP</acronym></cite> versions are installed on
|
|---|
| 257 | your machine (<abbr>e.g.</abbr>, one provided by the system in
|
|---|
| 258 | <samp>/usr/{include,lib}</samp> and a new one installed by the owner or
|
|---|
| 259 | administrator of the machine in <samp>/usr/local/{include,lib}</samp>)
|
|---|
| 260 | and your include and library search paths are inconsistent. On various
|
|---|
| 261 | <acronym>GNU</acronym>/Linux machines, this is unfortunately the case
|
|---|
| 262 | by default (<samp>/usr/local/include</samp> is in the default include
|
|---|
| 263 | search path, but <samp>/usr/local/lib</samp> is <em>not</em> in the
|
|---|
| 264 | default library search path). Typical errors are:
|
|---|
| 265 | <samp class="block-code">undefined reference to `__gmp_get_memory_functions'</samp>
|
|---|
| 266 | in <samp>make check</samp>. The best solution is to add
|
|---|
| 267 | <samp>/usr/local/include</samp> to your <var class="env">C_INCLUDE_PATH</var>
|
|---|
| 268 | environment variable and to add <samp>/usr/local/lib</samp> to your
|
|---|
| 269 | <var class="env">LIBRARY_PATH</var> and <var class="env">LD_LIBRARY_PATH</var>
|
|---|
| 270 | environment variables (and/or <var class="env">LD_RUN_PATH</var>).
|
|---|
| 271 | Alternatively, you can use <samp>--with-gmp*</samp> configure options,
|
|---|
| 272 | <abbr>e.g.</abbr> <samp>--with-gmp=/usr/local</samp>, but <strong>this is
|
|---|
| 273 | not guaranteed to work</strong> (in particular with <samp>gcc</samp> and
|
|---|
| 274 | system directories such as <samp>/usr</samp> or <samp>/usr/local</samp>),
|
|---|
| 275 | and other software that uses <cite><acronym>GMP</acronym></cite> and/or
|
|---|
| 276 | <cite><acronym>MPFR</acronym></cite> will need correct paths too;
|
|---|
| 277 | environment variables allow you to set them in a global way.</p>
|
|---|
| 278 | <p>Other information can be given in the <samp>INSTALL</samp> file and
|
|---|
| 279 | <samp>ld</samp> manual. Please look at them for more details. See also
|
|---|
| 280 | the <a href="#undef_ref2">next question</a>.</p></dd>
|
|---|
| 281 |
|
|---|
| 282 | <dt id="undef_ref2">5. When I link my program with
|
|---|
| 283 | <cite><acronym>MPFR</acronym></cite>, I get undefined reference
|
|---|
| 284 | to <code>__gmpXXXX</code>.</dt>
|
|---|
| 285 |
|
|---|
| 286 | <dd><p>Link your program with <cite><acronym>GMP</acronym></cite>. Assuming
|
|---|
| 287 | that your program is <samp>foo.c</samp>, you should link it using:
|
|---|
| 288 | <samp class="block-code">cc link.c -lmpfr -lgmp</samp>
|
|---|
| 289 | <cite><acronym>MPFR</acronym></cite> library reference (<samp>-lmpfr</samp>)
|
|---|
| 290 | should be before <cite><acronym>GMP</acronym></cite>'s one
|
|---|
| 291 | (<samp>-lgmp</samp>). Another solution is, with <acronym>GNU</acronym>
|
|---|
| 292 | <samp>ld</samp>, to give all the libraries inside a group:
|
|---|
| 293 | <samp class="block-code">gcc link.c -Wl,--start-group libgmp.a libmpfr.a -Wl,--end-group</samp>
|
|---|
| 294 | See <samp>INSTALL</samp> file and <samp>ld</samp> manual for more
|
|---|
| 295 | details.</p>
|
|---|
| 296 | <p>If you used correct link options, but still get an error, this may mean
|
|---|
| 297 | that your include and library search paths are inconsistent. Please see the
|
|---|
| 298 | <a href="#undef_ref1">previous question</a>.</p></dd>
|
|---|
| 299 |
|
|---|
| 300 | <dt id="crash_high_prec">6. My program crashes with high precisions.</dt>
|
|---|
| 301 |
|
|---|
| 302 | <dd><p>Your stack size limit may be too small; indeed, by default,
|
|---|
| 303 | <cite><acronym>GMP</acronym></cite> 4.1.4 and below allocates all
|
|---|
| 304 | temporary results on the stack, and in very high precisions, this
|
|---|
| 305 | limit may be reached. You can solve this problem in different ways:</p>
|
|---|
| 306 | <ul>
|
|---|
| 307 | <li><p>You can upgrade to <cite><acronym>GMP</acronym></cite> 4.2 (or above),
|
|---|
| 308 | which now makes temporary allocations on the stack only when they are
|
|---|
| 309 | small.</p></li>
|
|---|
| 310 | <li><p>You can increase the stack size limit with the <samp>limit</samp>,
|
|---|
| 311 | <samp>unlimit</samp> or <samp>ulimit</samp> command, depending on your
|
|---|
| 312 | shell. This may fail on some systems, where the maximum stack size cannot
|
|---|
| 313 | be increased above some value.</p></li>
|
|---|
| 314 | <li><p>You can rebuild both <cite><acronym>GMP</acronym></cite> and
|
|---|
| 315 | <cite><acronym>MPFR</acronym></cite> to use another allocation method.</p></li>
|
|---|
| 316 | </ul></dd>
|
|---|
| 317 |
|
|---|
| 318 | <dt id="accuracy">7. Though I have increased the precision, the results
|
|---|
| 319 | are not more accurate.</dt>
|
|---|
| 320 |
|
|---|
| 321 | <dd><p>The reason may be the use of C floating-point numbers. If you want
|
|---|
| 322 | to store a floating-point constant to a <code>mpfr_t</code>, you should use
|
|---|
| 323 | <code>mpfr_set_str</code> (or one of the <cite><acronym>MPFR</acronym></cite>
|
|---|
| 324 | constant functions, such as <code>mpfr_const_pi</code> for π) instead
|
|---|
| 325 | of <code>mpfr_set_d</code> or <code>mpfr_set_ld</code>. Otherwise the
|
|---|
| 326 | floating-point constant will be first converted into a reduced-precision
|
|---|
| 327 | (<abbr>e.g.</abbr>, 53-bit) binary number before
|
|---|
| 328 | <cite><acronym>MPFR</acronym></cite> can work with it. This is the case
|
|---|
| 329 | in particular for most exact decimal numbers, such as 0.17, which are
|
|---|
| 330 | not exactly representable in binary.</p>
|
|---|
| 331 | <p>Also remember that <cite><acronym>MPFR</acronym></cite> does not track
|
|---|
| 332 | the accuracy of the results: copying a value <var>x</var> to <var>y</var>
|
|---|
| 333 | with <code>mpfr_set (y, x, MPFR_RNDN)</code> where the variable <var>y</var>
|
|---|
| 334 | is more precise than the variable <var>x</var> will not make it more
|
|---|
| 335 | accurate; the (binary) value will remain unchanged.</p></dd>
|
|---|
| 336 |
|
|---|
| 337 | <dt id="detect_mpfr">8. How can I detect <cite><acronym>MPFR</acronym></cite>
|
|---|
| 338 | installation using <cite>autoconf</cite> or <cite>pkg-config</cite>?</dt>
|
|---|
| 339 |
|
|---|
| 340 | <dd><p>The <cite><acronym>MPFR</acronym></cite> team does not currently
|
|---|
| 341 | recommend any <cite>autoconf</cite> code, but a section will later
|
|---|
| 342 | be added to the <cite><acronym>MPFR</acronym></cite> manual.
|
|---|
| 343 | Limited <cite>pkg-config</cite> support has been added for
|
|---|
| 344 | <cite><acronym>MPFR</acronym></cite> 4.0.0; example:</p>
|
|---|
| 345 | <pre style="margin-left: 2em">cc myprogram.c $(pkg-config --cflags --libs mpfr)</pre></dd>
|
|---|
| 346 |
|
|---|
| 347 | <dt id="cite">9. How to cite <cite><acronym>MPFR</acronym></cite> in a
|
|---|
| 348 | scientific publication?</dt>
|
|---|
| 349 |
|
|---|
| 350 | <dd><p>To properly cite <cite><acronym>MPFR</acronym></cite> in a scientific
|
|---|
| 351 | publication, please cite the
|
|---|
| 352 | <a href="https://doi.org/10.1145/1236463.1236468"><acronym title="Association for Computing Machinery">ACM</acronym>
|
|---|
| 353 | <acronym title="Transactions on Mathematical Software">TOMS</acronym>
|
|---|
| 354 | paper</a>
|
|---|
| 355 | and/or the library web page
|
|---|
| 356 | <a href="https://www.mpfr.org/">https://www.mpfr.org</a>. If your publication
|
|---|
| 357 | is related to a particular release of <cite><acronym>MPFR</acronym></cite>,
|
|---|
| 358 | for example if you report timings, please also indicate the release number
|
|---|
| 359 | for future reference.</p></dd>
|
|---|
| 360 |
|
|---|
| 361 | <dt id="fpic">10. When I build <cite><acronym>MPFR</acronym></cite>, I get
|
|---|
| 362 | an error asking me to recompile with <samp>-fPIC</samp>.</dt>
|
|---|
| 363 |
|
|---|
| 364 | <dd><p>A typical error looks like:</p>
|
|---|
| 365 | <p><tt>/usr/bin/ld: <em>/path/to/</em>libgmp.a(realloc.o): relocation
|
|---|
| 366 | R_X86_64_32 against `.rodata.str1.1' can not be used when making a
|
|---|
| 367 | shared object; recompile with -fPIC<br />
|
|---|
| 368 | <em>/path/to/</em>libgmp.a: could not read symbols: Bad value<br />
|
|---|
| 369 | collect2: ld returned 1 exit status</tt></p>
|
|---|
| 370 | <p>The probable reason is that you tried to build
|
|---|
| 371 | <cite><acronym>MPFR</acronym></cite> with the shared library enabled (this
|
|---|
| 372 | is the default), while only a static <cite><acronym>GMP</acronym></cite>
|
|---|
| 373 | library could be found. To solve this problem, either rebuild and reinstall
|
|---|
| 374 | <cite><acronym>GMP</acronym></cite> without the <samp>--disable-shared</samp>
|
|---|
| 375 | configure option, or configure <cite><acronym>MPFR</acronym></cite> with
|
|---|
| 376 | <samp>--disable-shared</samp>. If you did this and still get the above
|
|---|
| 377 | error, the cause may be conflicting <cite><acronym>GMP</acronym></cite>
|
|---|
| 378 | versions installed on your system; please check that your search path
|
|---|
| 379 | settings are correct.</p>
|
|---|
| 380 | <p>Additional note about the last sentence: Under <acronym>GNU</acronym>/Linux
|
|---|
| 381 | (for instance), the linker takes the first library found in the library search
|
|---|
| 382 | path, whether it is dynamic or static. The default behavior under darwin is
|
|---|
| 383 | different, but <cite><acronym>MPFR</acronym></cite> will change it.</p></dd>
|
|---|
| 384 | <!-- Reference concerning darwin: see MPFR_LD_SEARCH_PATHS_FIRST
|
|---|
| 385 | in MPFR's configure.{ac,in} and acinclude.m4 -->
|
|---|
| 386 |
|
|---|
| 387 | </dl>
|
|---|
| 388 |
|
|---|
| 389 | </body>
|
|---|
| 390 |
|
|---|
| 391 | </html>
|
|---|
