INSTALL revision 1.1.1.5
1 Copyright 1999-2020 Free Software Foundation, Inc. 2 Contributed by the AriC and Caramba 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 https://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 Installing GNU MPFR 24 =================== 25 26 Note: In case of problem, please read this INSTALL file carefully before 27 reporting a bug, in particular Section "In case of problem" below. Some 28 problems are due to bad configuration on the user side (not specific to 29 MPFR). 30 31 0. You first need to install GMP. See <https://gmplib.org/>. 32 MPFR requires GMP version 5.0.0 or later. 33 34 1. Extract the files from the archive. 35 36 2. It is strongly advised to apply the latest patches if this has 37 not been done yet and if patches are available. You can check 38 on the release page: 39 40 https://www.mpfr.org/mpfr-4.1.0/ 41 42 which may have additional information. The patches can be applied 43 with commands like: 44 45 wget https://www.mpfr.org/mpfr-4.1.0/allpatches 46 patch -N -Z -p1 < allpatches 47 48 or 49 50 curl https://www.mpfr.org/mpfr-4.1.0/allpatches | patch -N -Z -p1 51 52 (Those instructions are for the GNU patch command, for example 53 /usr/bin/gpatch on Solaris.) 54 55 3. In the MPFR directory, to detect your system, type: 56 ./configure 57 possibly with options (see below, in particular if this step or 58 one of the following fails). You should also check whether WARNING 59 lines have been output (such a problem may cause a failure in one 60 of the following steps). 61 Note: paths provided in configure options must always be absolute 62 (relative paths are not supported). 63 64 4. To build the library, type: 65 make 66 67 [optional] if you want to tune MPFR for your specific architecture, see 68 the section "Tuning MPFR" below. Note that for most common architectures, 69 MPFR includes some default tuning parameters which should be near from 70 optimal. 71 72 5. To check the built library (runs the test files), type: 73 make check 74 75 Note: If any test fails, information about this failure can be found in 76 the tests/test-suite.log file; you should provide this file in your bug 77 reports (in addition to other useful information, as mentioned later). 78 If you want the contents of this file to be automatically output in case 79 of failure, you can set the VERBOSE environment variable to 1 before 80 running "make check", for instance by typing: 81 VERBOSE=1 make check 82 83 6. [Optional / experimental] Binary distributions may also want to run: 84 make check-gmp-symbols 85 86 This will check that MPFR does not use GMP internal symbols, which 87 could yield failures in case of GMP upgrade without a MPFR rebuild. 88 But note that this is a heuristic and might give false positives or 89 false negatives. Please report any problem to the MPFR developers. 90 End users may also be interested in this check, unless they have 91 allowed GMP internals with configure options (see below). 92 93 7. To install it (default "/usr/local" | see "--prefix" option), type: 94 make install 95 96 If you installed MPFR (header and library) in directories that are 97 not searched by default by the compiler and/or linking tools, then, 98 like with other libraries, you may need to set up some environment 99 variables such as C_INCLUDE_PATH (to find the header mpfr.h), 100 LIBRARY_PATH (to find the library), and if the shared library has 101 been installed, LD_LIBRARY_PATH (before execution) or LD_RUN_PATH 102 (before linking); this list is not exhaustive and some environment 103 variables may be specific to your system. "make install" gives some 104 instructions; please read them. You can also find more information 105 in the manuals of your compiler and linker. The MPFR FAQ may also 106 give some information. 107 108 Remember that if you have several MPFR (or GMP) versions installed 109 (e.g., one with the system, and one, newer, by you), you will not 110 necessarily get a compilation/linking error if a wrong library is 111 used (e.g., because LD_LIBRARY_PATH has not been set correctly). 112 But unexpected results may occur. 113 114 Under Mac OS X, if the shared library was not installed and you use 115 Apple's linker (this is the default), you will also need to provide 116 the -search_paths_first linker flag ("-Wl,-search_paths_first" when 117 you link via gcc) to make sure that the right library is selected, 118 as by default, Apple's linker selects a shared library preferably, 119 even when it is farther in the library paths. We recall that if a 120 wrong library is selected due to this behavior, unexpected results 121 may occur. 122 123 124 Building the documentation 125 ========================== 126 127 To build the documentation in various formats, you may first need to 128 install recent versions of some utilities such as texinfo. 129 130 * Type "make info" to produce the documentation in the info format. 131 132 * Type "make pdf" to produce the documentation in the PDF format. 133 134 * Type "make dvi" to produce the documentation in the DVI format. 135 136 * Type "make ps" to produce the documentation in the Postscript format. 137 138 * Type "make html" to produce the documentation in the HTML format 139 (in several pages); if you want only one output HTML file, then 140 type "makeinfo --html --no-split mpfr.texi" from the doc directory 141 instead. 142 143 144 Building MPFR with internal GMP header files 145 ============================================ 146 147 MPFR built with internal GMP header files is a bit faster, so you may want 148 to build it with them. Just do this in step 1: 149 150 ./configure --with-gmp-build=GMPBUILD 151 152 where GMPBUILD is the GMP build directory. The needed header files are: 153 gmp-impl.h, longlong.h and all the necessary headers to use them, which 154 may be located either in the GMP source directory or in the GMP build 155 directory, in case they are different (MPFR takes care of that, as of 156 MPFR 3.1.0). 157 158 Warning: the library obtained in this way may use some internal GMP 159 symbols, and thus dynamically linking your software with a different 160 version of GMP might fail, even though it is declared as compatible 161 by Libtool's versioning system. 162 163 164 Tuning MPFR 165 =========== 166 167 [For the current GMP version (6.1.0), a Unix-like OS is required.] 168 169 For this, you need to build MPFR with a GMP build directory (see above). 170 In the GMP build directory, you also need to go into the "tune" subdirectory 171 and type "make speed". This will build the GMP speed library, which is used 172 by the MPFR tuning mechanism. 173 174 Then go back to the MPFR build directory, go into the "tune" subdirectory and 175 type "make tune". This will build an optimized file "mparam.h" for your 176 specific architecture. 177 178 179 ./configure options 180 =================== 181 182 --prefix=DIR installs MPFR headers and library in DIR/include and 183 DIR/lib respectively (the default is "/usr/local"). 184 185 --with-gmp-include=DIR assumes that DIR contains gmp.h 186 --with-gmp-lib=DIR assumes that DIR contains the GMP library 187 --with-gmp=DIR assumes that DIR is where you have installed GMP. 188 same as --with-gmp-lib=DIR/lib 189 and --with-gmp-include=DIR/include 190 (use either --with-gmp alone or one or both of 191 --with-gmp-lib/--with-gmp-include) 192 Warning! Do not use these options if you have 193 CPPFLAGS and/or LDFLAGS containing a -I or -L 194 option with a directory that contains a GMP 195 header or library file, as these options just 196 add -I and -L options to CPPFLAGS and LDFLAGS 197 *after* the ones that are currently declared, 198 so that DIR will have a lower precedence. Also, 199 this may not work if DIR is a system directory 200 (typically /usr or /usr/local); see below. 201 202 --with-gmp-build=DIR assumes that DIR contains the GMP build directory, 203 and enables the use of GMP internals (see above). 204 Warning! This option and the group of options 205 --with-gmp are mutually exclusive. 206 207 --enable-assert build MPFR with assertions. 208 209 --enable-thread-safe build MPFR as thread safe, using compiler-level 210 Thread-Local Storage (TLS). Note: TLS support is 211 roughly tested by configure. If configure detects 212 that TLS does not work (because of the compiler, 213 linker or system libraries), it will output an 214 error message, telling you to build MPFR without 215 thread safe. For instance, though Mac OS X uses 216 GCC, it may not currently support GCC's __thread 217 storage class. 218 219 --disable-thread-safe build MPFR without TLS. By default, TLS support 220 is detected automatically, and MPFR is built as 221 thread safe if supported. However this detection 222 is only a heuristic: TLS can be detected as 223 supported while its support is incomplete or 224 buggy (MPFR tests may fail). In such a case, 225 this option is useful. 226 227 --enable-decimal-float build conversion functions from/to decimal floats. 228 Note that detection by the configure script is 229 limited in case of cross-compilation. 230 Accepted arguments: 231 yes Decimal support is requested and the configure 232 script fails if it detects that decimals do not 233 work. 234 The encoding (BID or DPD) will automatically be 235 detected at configure time or at compile time if 236 possible (if not, generic code will be used). 237 no Decimal support is explicitly disabled. 238 auto Decimal support is enabled if the configure script 239 detects that it works. This is the default when 240 --{enable,disable}-decimal-float is not given. 241 bid Decimal support is requested and the encoding is 242 assumed to be BID (some check may be done). 243 dpd Decimal support is requested and the encoding is 244 assumed to be DPD (some check may be done). 245 generic Decimal support is requested and the generic code 246 is used (mainly for developers). 247 248 --enable-gmp-internals allows the MPFR build to use GMP's undocumented 249 functions (not from the public API). Note that 250 library versioning is not guaranteed to work if 251 this option is used. Thus it must not be used in 252 binary distributions. 253 254 --with-sysroot=DIR Search for dependent libraries within DIR (which 255 may be useful in cross-compilation). If you use 256 this option, you need to have Libtool 2.4+ on 257 the target system. See Libtool 2.4+'s NEWS file. 258 259 Note: By default, the configure script tries to set CC / CFLAGS to GMP's 260 ones from gmp.h (__GMP_CC / __GMP_CFLAGS) in order to ensure that MPFR is 261 built with the same ABI as GMP. The reason is that when GMP is built, it 262 may set CC / CFLAGS to select an ABI that is not the default one in order 263 to have a better performance. The -pedantic option in GMP's CFLAGS, when 264 present (which is the case by default), is removed, because the MPFR 265 build system uses some C extensions (when this script detects that they 266 are supported) and -pedantic yields too many useless warnings. However, 267 this setting from GMP is not guaranteed to work as the configure script 268 does some compiler tests earlier, and a conflict may arise. Also, the 269 values obtained from GMP may be incorrect for the MPFR build if GMP has 270 been built on a different machine; in such a case, the user may need to 271 specify CC / CFLAGS, as explained below. 272 273 Moreover, even without --with-gmp-build and --enable-gmp-internals, 274 MPFR might use some GMP internals by mistake. This would be a bug, 275 which should be reported to the MPFR developers. 276 277 Run "./configure --help" to see the other options (default options 278 from Autoconf and Automake). 279 280 281 If 'gmp.h' and 'libgmp' do not match 282 ==================================== 283 284 Under some circumstances, the configure script may output a message 285 saying: 286 287 'gmp.h' and 'libgmp' seem to have different versions or 288 we cannot run a program linked with GMP (if you cannot 289 see the version numbers above). [...] 290 291 Even though the configure script does not fail in such a case, this 292 message most often indicates a real error, which needs to be avoided. 293 Possible causes are: 294 295 * The wanted GMP library does not have the same ABI as the one chosen 296 to build MPFR. The consequences may be: 297 298 - A different GMP library (with the correct ABI) has been found, 299 but does not have the same version as 'gmp.h'. 300 301 - No other GMP libraries have been found (in this case, no version 302 numbers should be printed above the warning message). 303 304 This is incorrect and one of the following steps (make, make check) 305 will probably fail. GMP (actually gmp.h) now provides CC and CFLAGS 306 information to select the correct ABI, so that this problem should 307 no longer occur; but if GMP was provided by a binary package, such 308 information may not be valid. See the 309 310 checking for CC and CFLAGS in gmp.h... 311 312 line in the configure output (about the 11th line) and the following 313 few ones for more information. You may need to reinstall GMP or to 314 provide your own CC and/or CFLAGS. See also the remaining of this 315 INSTALL file. 316 317 * A configure option like --with-gmp or --with-gmp-include was used 318 with a system include directory, e.g. one of the following: 319 320 --with-gmp=/usr 321 --with-gmp=/usr/local 322 --with-gmp-include=/usr/include 323 --with-gmp-include=/usr/local/include 324 325 GCC (and possibly other compilers) will ignore such a directive for 326 include directories (but this rule is not applied for the library 327 itself!). This means that the library search paths will be reordered 328 as declared, but the specified include directory will still be near 329 the end of the include search paths (thus with a low precedence). 330 This is not a problem if only one GMP version is installed, but 331 otherwise, a wrong gmp.h may be chosen, so that the versions of 332 gmp.h and libgmp may not match. The suggestions are the following: 333 334 - If you want to use the GMP version under /usr, then you should 335 uninstall all the other GMP versions (header and library files) 336 that may be seen in the search paths, in particular those under 337 /usr/local. 338 339 - If you want to use the GMP version under /usr/local, then you 340 should uninstall all the other GMP versions (header and library 341 files) that may be seen in the search paths, but *NOT* the one 342 under /usr (the version under /usr is provided by the OS vendor, 343 and changing/removing anything related to it may break your 344 system, and /usr should have a lower precedence than /usr/local 345 anyway). 346 347 To find where GMP versions have been installed: 348 349 $ locate libgmp (if you have a locate database) 350 351 and if the compiler is GCC: 352 353 $ gcc -print-file-name=libgmp.so (under most systems) 354 $ gcc -print-file-name=libgmp.dylib (under Mac OS X) 355 356 and if this does not work, you may try: 357 358 $ gcc -print-search-dirs 359 360 * An official GCC version was used under Debian GNU/Linux. Problems 361 may come from the fact that Debian chose a different convention 362 for library search paths concerning 32-bit and 64-bit libraries. 363 A possible problem can be, for instance: 364 365 [Debian's GCC, correct library path] 366 $ gcc -print-file-name=libgmp.so 367 /home/vlefevre/gmp/athlon64/lib/../lib/libgmp.so 368 369 [Official GCC, incorrect library path] 370 $ gcc-4.3.1 -print-file-name=libgmp.so 371 /usr/lib/../lib64/libgmp.so 372 373 The solution: use a GCC provided by Debian or add symbolic links 374 such as lib64 -> lib (on 64-bit machines) for your library paths. 375 376 * The problem may also be temporary and only due to the fact that 377 libtool could not be used at this time. This is unlikely, though. 378 379 380 In case of problem 381 ================== 382 383 First, look for any warning message in the configure output. 384 385 Several documents may help you to solve the problem: 386 * this INSTALL file, in particular information given below; 387 * the FAQ (either the FAQ.html file distributed with MPFR, or the 388 on-line version <https://www.mpfr.org/faq.html>, which may be more 389 up-to-date); 390 * the MPFR web page for this version <https://www.mpfr.org/mpfr-4.1.0/>, 391 which lists bugs found in this version and provides some patches. 392 393 If the "configure" fails, please check that the C compiler and its 394 options are the same as those for the GMP build (specially the ABI). 395 You can see the latter with the following command: 396 397 grep "^CC\|^CFLAGS" GMPBUILD/Makefile 398 399 if the GMP build directory is available. Then type: 400 401 ./configure <configure options> CC=<C compiler> CFLAGS="<compiler options>" 402 403 (quotes are needed when there are spaces or other special characters 404 in the CC/CFLAGS value) and continue the install. On some platforms, 405 you should provide further options to match those used by GMP, or set 406 some environment variables. For instance, see the "Notes on AIX/PowerPC" 407 section below. 408 409 Warning! Do NOT use optimization options that can change the semantics 410 of math operations, such as GCC's -ffast-math or Sun CC's -fast. 411 Otherwise conversions from/to double's may be incorrect on infinities, 412 NaN's and signed zeros. Since native FP arithmetic is used in a few 413 places only, such options would not make MPFR faster anyway. 414 415 If some "make check" tests fail, you can try the --disable-thread-safe 416 configure option (see the configure options above): it has been reported 417 that some platforms have buggy TLS support. Before trying this option, 418 you may want to check in the configure output whether MPFR was built 419 with TLS support; if yes, you will have a line: 420 checking for TLS support... yes 421 Alternatively "grep MPFR_USE_THREAD_SAFE config.log" will show that 422 MPFR_USE_THREAD_SAFE is defined to 1. If it is "no" (or the variable 423 is not defined), the --disable-thread-safe option would be useless. 424 425 Some tests failure may be due to other compiler bugs, in particular 426 in optimization code. You can try to build MPFR without compiler 427 optimizations by giving -O0 (letter O, digit 0) in CFLAGS. If the 428 MPFR tests no longer fail, this was probably due to a compiler bug, 429 though we cannot exclude a bug in MPFR. You may want to contact us 430 (see below), possibly after looking at: 431 https://members.loria.fr/PZimmermann/software/compilerbugs.html 432 433 On some platforms, try with "gmake" (GNU make) instead of "make". 434 Problems have been reported with the Tru64 make. 435 436 If the configure script reports that gmp.h version and libgmp version 437 are different, or if the build was OK, but the tests failed to link 438 with GMP or gave an error like 439 440 undefined reference to `__gmp_get_memory_functions' 441 442 meaning that the GMP library was not found or a wrong GMP library was 443 selected by the linker, then your library search paths are probably 444 not correctly set (some paths are missing or they are specified in an 445 incorrect order). 446 447 Such problems commonly occur under some GNU/Linux machines, where the 448 default header and library search paths may be inconsistent: GCC is 449 configured to search /usr/local/include and /usr/local/lib by default, 450 while the dynamic linker ignores /usr/local/lib. If you have a GMP 451 version installed in /usr (provided by the OS vendor) and a new one 452 installed in /usr/local, then the header of the new GMP version and 453 the library of the old GMP version will be used! The best solution 454 is to make sure that the dynamic linker configuration is consistent 455 with GCC's behavior, for instance by having /usr/local/lib in 456 /etc/ld.so.conf or in some file from /etc/ld.so.conf.d (as Debian 457 did: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=395177). See 458 also https://gcc.gnu.org/legacy-ml/gcc-help/2010-01/msg00171.html for 459 more information. Alternatively you can use: 460 * environment variables. This may sometimes be necessary. If DIR 461 is the installation directory of GMP, add DIR/include to your 462 CPATH or C_INCLUDE_PATH (for compilers other than GCC, please 463 check the manual of your compiler), and add DIR/lib to your 464 LIBRARY_PATH and LD_LIBRARY_PATH (and/or LD_RUN_PATH); 465 * --with-gmp* configure options (described above), e.g. 466 --with-gmp=/opt/local (to use /opt/local/include for headers and 467 /opt/local/lib for libraries), but other software that uses GMP 468 and/or MPFR will need correct paths too, and environment variables 469 allow one to set such search paths in a global way. 470 Note about "--with-gmp=/usr/local". This option may appear to 471 solve the above inconsistency problem, but does not work as you 472 expect. Indeed it affects the library search path, in particular, 473 the one used by the dynamic linker (thus adding the missing 474 /usr/local/lib directory as wanted), but since /usr/local/include 475 is a "standard system include directory" for GCC, the include 476 search patch is not changed; this is often not a problem in this 477 particular case because usually, /usr/local/include is already 478 last in the include search patch, but this may fail under some 479 occasions and may trigger obscure errors. 480 481 For instance, under Unix, where paths are separated by a colon: 482 483 * With POSIX sh-compatible shells (e.g. sh, ksh, bash, zsh): 484 export C_INCLUDE_PATH="/usr/local/include:/other/path/include" 485 export LIBRARY_PATH="/usr/local/lib:/other/path/lib" 486 export LD_LIBRARY_PATH="$LIBRARY_PATH" 487 488 * With csh or tcsh: 489 setenv C_INCLUDE_PATH "/usr/local/include:/other/path/include" 490 setenv LIBRARY_PATH "/usr/local/lib:/other/path/lib" 491 setenv LD_LIBRARY_PATH "$LIBRARY_PATH" 492 493 If almost all the tests fail and the messages in the test-suite.log file 494 (or in the output, when running individual tests from the command line) 495 start with a line of the form: 496 497 Incorrect MPFR version! (xxx header vs yyy library) 498 499 then this means that an installed MPFR version is tested instead of the 500 one that has just been built. This is probably not a bug in MPFR, but a 501 problem caused by the user or system configuration (particular options, 502 environment variables, etc.) or a bug in the toolchain. In particular, 503 if LD_LIBRARY_PATH overrides the run path (set up by libtool) and an 504 installed ABI-compatible version of MPFR is in a directory listed in 505 the LD_LIBRARY_PATH search path, then this will break. An example with 506 GNU ld: 507 508 https://sourceware.org/bugzilla/show_bug.cgi?id=21476 509 510 If you can't solve your problem, you should contact us via the MPFR 511 mailing-list <mpfr (a] inria.fr>, indicating the machine and operating system 512 used (uname -a), the compiler and version used (gcc -v if you use gcc), 513 the configure options used if any (including variables such as CC and 514 CFLAGS), the version of GMP and MPFR used, and a description of the 515 problem encountered, in particular the tests/test-suite.log file if 516 "make check" failed. Please send us also the log of the "configure" 517 (config.log). 518 519 520 Notes about ABI 521 =============== 522 523 On 64-bit computers, it may happen that the default ABI (Application Binary 524 Interface) chosen by MPFR does not correspond to the default one chosen by 525 the compiler. 526 527 In particular, this kind of message may indicate the problem: 528 /usr/bin/ld: skipping incompatible mpfr/src/.libs/libmpfr.a when searching for -lmpfr 529 530 In fact, since MPFR relies on GMP, it uses the same ABI as GMP. 531 To be sure that your program compiles and links correctly, use the same 532 compiler flags as MPFR does (look for CFLAGS in config.log). 533 534 You might also recompile GMP with a different ABI, with for example 535 ./configure ABI=32. 536 537 538 Notes on Mac OS X 539 ================= 540 541 If you get an error of the form 542 543 ld: pointer in read-only segment not allowed in slidable image... 544 545 this can mean that the link is done against a static (GMP) library. 546 In such a case, you should configure MPFR with --disable-shared to 547 disable the build of the shared library. 548 549 550 Notes on FreeBSD 4.3 551 ==================== 552 553 FreeBSD 4.3 is provided with an incorrect <float.h> header file, and 554 MPFR tests related to long double's may fail. If you cannot upgrade 555 the system, you can still use MPFR with FreeBSD 4.3, but you should 556 not use conversions with the long double type. 557 558 559 Notes on AIX/PowerPC 560 ==================== 561 562 The following has been tested on AIX 7.1.3 (gcc111.fsffrance.org) 563 with gcc 4.8.1 and GMP 6.1.0. 564 565 If GMP was built with the 64-bit ABI, before building and testing MPFR, 566 you should set the OBJECT_MODE environment variable to 64, e.g., with: 567 568 export OBJECT_MODE=64 569 570 (in a sh-compatible shell). Alternatively add the following to the configure 571 line: AR="ar -X64" NM="nm -B -X64". 572 573 574 Notes on Solaris 575 ================ 576 577 Do not put a -R<dir> option in the LD_OPTIONS environment variable, at least 578 if the directory <dir> can contain an MPFR library. Otherwise this MPFR 579 library may be chosen for the tests (make check) instead of the one that has 580 just been built, in which case, either you will get errors due to unmatched 581 versions or this problem may remain undetected. The reason is that this 582 option will appear before the -R options added by libtool, such as the one 583 to the src/.libs directory containing the MPFR library that has just been 584 built, and will have the precedence. 585 586 587 MPFR for use with Windows Applications 588 ====================================== 589 590 There are several ways of building MPFR on Windows, the most appropriate 591 approach depending on how you intend to use the resulting libraries. 592 593 a. Using MinGW 594 ============== 595 596 1 - We advise to use MinGW (http://www.mingw.org/ for 32-bit, and 597 https://sourceforge.net/projects/mingw-w64/ for 32- and 64-bit), 598 which is simpler and less demanding than Cygwin. Contrary to Cygwin, 599 it also provides native Windows code. 600 601 If you also use MSYS, you should use "make" for MSYS instead of 602 the "make" utility from MinGW-W64 or from GCC, which causes the 603 following error: 604 605 libtool: warning: libobj name 'extract.Tpo -c -o extract.lo extract.lo' 606 may not contain shell special characters. 607 rm: unknown option -- c 608 609 References about this issue and solution: 610 https://sourceforge.net/p/msys2/tickets/223/ 611 https://sympa.inria.fr/sympa/arc/mpfr/2016-07/msg00009.html 612 613 2 - If you just want to make a binary with gcc, there is nothing to do: 614 GMP, MPFR and the program compile exactly as under Linux. (It is 615 recommended to pass --build=xxx-yyy-mingw64 to the GMP configure command, 616 or --build=xxx with xxx containing mingw.) 617 618 3 - To avoid using the Microsoft runtime (which might not be conform to ISO C), 619 you can use the MinGW runtime package (which is an integral part of MinGW). 620 621 With old MinGW versions, you can get an ISO-compliant printf() 622 if you compile your application with either '-ansi', '-posix' or 623 '-D__USE_MINGW_ANSI_STDIO'. But note that this latter option, 624 which was useful in the past (see below) should no longer be used. 625 626 The following applies to old MinGW versions, and may be discouraged 627 with recent MinGW versions. 628 629 In order to have the MPFR formatted output functions based on an 630 ISO-compliant printf(), you need to compile GMP (not MPFR) with 631 CC="gcc -D__USE_MINGW_ANSI_STDIO" (since the standard printf modifiers 632 %e, %Ld and %td are passed to GMP). 633 634 To avoid failures of some of the printf-related tests, MPFR needs to 635 be compiled with the same __USE_MINGW_ANSI_STDIO as with GMP, i.e. 636 this macro should be defined for both or undefined for both; this 637 should be the case by default, unless CC or CFLAGS has been redefined. 638 For instance, if __USE_MINGW_ANSI_STDIO is defined in GMP but not in 639 MPFR, the following error may occur: 640 641 Error in mpfr_vsprintf (s, "%e", ...); 642 expected: "-1.250000e+000" 643 got: "-1.250000e+00" 644 FAIL tsprintf.exe (exit status: 1) 645 646 and in the opposite case: 647 648 Error in mpfr_vsprintf (s, "%e", ...); 649 expected: "-1.250000e+00" 650 got: "-1.250000e+000" 651 FAIL tsprintf.exe (exit status: 1) 652 653 Note with old GMP versions: Other issues could arise due to the fact 654 that the C functions vsnprintf and vsprintf both used internally in 655 old GMP versions do not produce the same output: 656 657 https://sympa.inria.fr/sympa/arc/mpfr/2016-03/msg00045.html 658 https://sympa.inria.fr/sympa/arc/mpfr/2016-03/msg00051.html 659 https://sympa.inria.fr/sympa/arc/mpfr/2016-03/msg00053.html 660 661 If __USE_MINGW_ANSI_STDIO has not been defined when building GMP, 662 then the length modifiers j, L and t are not supported with the GMP 663 formatted output functions, and as a consequence, also with MPFR. 664 This is automatically detected by the configure script, except when 665 cross-compiling (e.g. under Linux), in which case some macros need 666 to be defined explicitly, e.g. with 667 668 "CPPFLAGS=-DNPRINTF_J -DNPRINTF_L -DNPRINTF_T" 669 670 in order to avoid potential issues with the MPFR library and failures 671 in the test suite (the corresponding tests are disabled explicitly). 672 673 b. Using Cygwin 674 =============== 675 676 This build should be similar to that for MinGW except that the resulting 677 library depends on the Cygwin DLL and cannot therefore be used with 678 Visual Studio as with MinGW. Indeed, the binaries compiled with Cygwin 679 require a dynamic library (cygwin.dll) to work; there is a Cygwin option 680 -mno-cygwin to build native code, but it may require some non-portable tricks. 681 682 In case of failure, you may need to pass LDFLAGS='-shared-libgcc' at the 683 end of the configure line due to a bug in GCC. Otherwise, if threading 684 support is not needed, you can configure MPFR with --disable-thread-safe. 685 686 c. Using Microsoft Visual C++ and Intel C++ Compilers 687 ===================================================== 688 689 Static and dynamic MPFR libraries for the 32- and 64-bit versions of 690 Windows can be built with Microsoft Visual Studio 2015 using the 691 Microsoft Visual C++ compiler, see: 692 693 https://visualstudio.microsoft.com/ 694 695 The Intel C++ compiler provided as a part of Intel Parallel Studio XE 696 can also be used: 697 698 https://software.intel.com/en-us/parallel-studio-xe 699 700 The relevant build projects are available here: 701 702 https://github.com/BrianGladman 703 704 d. Using the CompCert compiler 705 ============================== 706 707 [this was tested with CompCert 3.7 and MPFR revision 13851 on x86_64-linux] 708 709 CompCert (http://compcert.inria.fr/) is a formally verified compiler. 710 To compile MPFR with CompCert: 711 712 $ ./configure CC=ccomp CFLAGS="-fbitfields -flongdouble -fstruct-passing" --disable-shared 713 714 You also need to unset LD_LIBRARY_PATH, and/or you might need to change 715 wl="" into wl="-Wl," in the libtool file. 716 717 All tests (make check) should pass (tget_set_d64, tget_set_d128 and 718 tset_float128 are skipped, since CompCert does not support _Decimal64, 719 _Decimal128 nor _Float128). 720
Indexes created Tue Oct 27 00:25:24 UTC 2026