5101dcfb15
* gnu/packages/multiprecision.scm (qd)[arguments]: Add "--disable-static" to #:configure-flags.
519 lines
21 KiB
Scheme
519 lines
21 KiB
Scheme
;;; GNU Guix --- Functional package management for GNU
|
|
;;; Copyright © 2012, 2013, 2015 Ludovic Courtès <ludo@gnu.org>
|
|
;;; Copyright © 2014 Mark H Weaver <mhw@netris.org>
|
|
;;; Copyright © 2015, 2018 Andreas Enge <andreas@enge.fr>
|
|
;;; Copyright © 2016, 2020 Nicolas Goaziou <mail@nicolasgoaziou.fr>
|
|
;;; Copyright © 2016, 2020 Jan (janneke) Nieuwenhuizen <janneke@gnu.org>
|
|
;;; Copyright © 2018, 2019 Tobias Geerinckx-Rice <me@tobias.gr>
|
|
;;; Copyright © 2018 Eric Bavier <bavier@member.fsf.org>
|
|
;;; Copyright © 2018, 2019, 2021 Efraim Flashner <efraim@flashner.co.il>
|
|
;;; Copyright © 2021 Vinicius Monego <monego@posteo.net>
|
|
;;;
|
|
;;; This file is part of GNU Guix.
|
|
;;;
|
|
;;; GNU Guix is free software; you can redistribute it and/or modify it
|
|
;;; under the terms of the GNU General Public License as published by
|
|
;;; the Free Software Foundation; either version 3 of the License, or (at
|
|
;;; your option) any later version.
|
|
;;;
|
|
;;; GNU Guix 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 General Public License for more details.
|
|
;;;
|
|
;;; You should have received a copy of the GNU General Public License
|
|
;;; along with GNU Guix. If not, see <http://www.gnu.org/licenses/>.
|
|
|
|
(define-module (gnu packages multiprecision)
|
|
#:use-module (guix licenses)
|
|
#:use-module (gnu packages)
|
|
#:use-module (gnu packages autotools)
|
|
#:use-module (gnu packages m4)
|
|
#:use-module (gnu packages gcc)
|
|
#:use-module (gnu packages texinfo)
|
|
#:use-module (guix packages)
|
|
#:use-module (guix download)
|
|
#:use-module (guix utils)
|
|
#:use-module (guix build-system gnu))
|
|
|
|
(define-public gmp
|
|
(package
|
|
(name "gmp")
|
|
(version "6.2.1")
|
|
(source (origin
|
|
(method url-fetch)
|
|
(uri
|
|
(string-append "mirror://gnu/gmp/gmp-"
|
|
version ".tar.xz"))
|
|
(sha256
|
|
(base32
|
|
"1wml97fdmpcynsbw9yl77rj29qibfp652d0w3222zlfx5j8jjj7x"))
|
|
(patches (search-patches "gmp-faulty-test.patch"))))
|
|
(build-system gnu-build-system)
|
|
(native-inputs (list m4))
|
|
(outputs '("out" "debug"))
|
|
(arguments
|
|
`(#:parallel-tests? #f ; mpz/reuse fails otherwise
|
|
#:configure-flags
|
|
'(;; Build a "fat binary", with routines for several
|
|
;; sub-architectures.
|
|
"--enable-fat"
|
|
"--enable-cxx"
|
|
,@(cond ((target-mingw?)
|
|
;; Static and shared cannot be built in one go:
|
|
;; they produce different headers. We need shared.
|
|
`("--disable-static"
|
|
"--enable-shared"))
|
|
(else '())))
|
|
;; Remove after core-updates merge.
|
|
;; Workaround for gcc-7 transition breakage, -system and cross-build,
|
|
;; Note: See <http://bugs.gnu.org/22186> for why not 'CPATH'.
|
|
;; Note: See <http://bugs.gnu.org/30756> for why not 'C_INCLUDE_PATH' & co.
|
|
,@(if (target-mingw?)
|
|
`(#:phases
|
|
(modify-phases %standard-phases
|
|
(add-before 'configure 'setenv
|
|
(lambda _
|
|
(let ((gcc (assoc-ref %build-inputs "cross-gcc"))
|
|
(libc (assoc-ref %build-inputs "cross-libc")))
|
|
(setenv "CROSS_CPLUS_INCLUDE_PATH"
|
|
(string-append gcc "/include/c++"
|
|
":" gcc "/include"
|
|
":" libc "/include"))
|
|
(format #t "environment variable `CROSS_CPLUS_INCLUDE_PATH' set to `~a'\n"
|
|
(getenv "CROSS_CPLUS_INCLUDE_PATH"))
|
|
#t)))))
|
|
'())))
|
|
(synopsis "Multiple-precision arithmetic library")
|
|
(description
|
|
"The @acronym{GMP, the GNU Multiple Precision Arithmetic} library performs
|
|
arbitrary-precision arithmetic on signed integers, rational numbers and floating
|
|
point numbers. The precision is only limited by the available memory.
|
|
The library is highly optimized, with a design focus on execution speed.
|
|
It is aimed at use in, for example, cryptography and computational algebra.")
|
|
(license lgpl3+)
|
|
(home-page "https://gmplib.org/")))
|
|
|
|
(define-public gmp-6.0
|
|
;; We keep this one around to bootstrap GCC, to work around a compilation
|
|
;; issue on ARM. See
|
|
;; <https://gmplib.org/list-archives/gmp-bugs/2015-December/003848.html>.
|
|
(package
|
|
(inherit gmp)
|
|
(version "6.0.0a")
|
|
(source (origin
|
|
(method url-fetch)
|
|
(uri (string-append "mirror://gnu/gmp/gmp-"
|
|
version ".tar.xz"))
|
|
(sha256
|
|
(base32
|
|
"0r5pp27cy7ch3dg5v0rsny8bib1zfvrza6027g2mp5f6v8pd6mli"))
|
|
(patches (search-patches "gmp-arm-asm-nothumb.patch"
|
|
"gmp-faulty-test.patch"))))))
|
|
|
|
(define-public mpfr
|
|
(package
|
|
(name "mpfr")
|
|
(version "4.1.0")
|
|
(source (origin
|
|
(method url-fetch)
|
|
(uri (string-append "mirror://gnu/mpfr/mpfr-" version
|
|
".tar.xz"))
|
|
(sha256 (base32
|
|
"0zwaanakrqjf84lfr5hfsdr7hncwv9wj0mchlr7cmxigfgqs760c"))))
|
|
(build-system gnu-build-system)
|
|
(outputs '("out" "debug"))
|
|
(propagated-inputs (list gmp)) ; <mpfr.h> refers to <gmp.h>
|
|
(synopsis "C library for arbitrary-precision floating-point arithmetic")
|
|
(description
|
|
"GNU@tie{}@acronym{MPFR, Multiple Precision Floating-Point Reliably} is a C
|
|
library for performing multiple-precision, floating-point computations with
|
|
correct rounding.")
|
|
(license lgpl3+)
|
|
(home-page "https://www.mpfr.org/")))
|
|
|
|
(define-public mpc
|
|
(package
|
|
(name "mpc")
|
|
(version "1.2.1")
|
|
(source (origin
|
|
(method url-fetch)
|
|
(uri (string-append
|
|
"mirror://gnu/mpc/mpc-" version ".tar.gz"))
|
|
(sha256
|
|
(base32
|
|
"0n846hqfqvmsmim7qdlms0qr86f1hck19p12nq3g3z2x74n3sl0p"))))
|
|
(build-system gnu-build-system)
|
|
(outputs '("out" "debug"))
|
|
(propagated-inputs (list gmp ; <mpc.h> refers to both
|
|
mpfr))
|
|
(synopsis "C library for arbitrary-precision complex arithmetic")
|
|
(description
|
|
"GNU@tie{}@acronym{MPC, Multiple Precision Complex library} is a C library
|
|
for performing arithmetic on complex numbers. It supports arbitrarily high
|
|
precision and correctly rounds the results.")
|
|
(license lgpl3+)
|
|
(home-page "http://www.multiprecision.org/mpc/")))
|
|
|
|
(define-public mpfi
|
|
(package
|
|
(name "mpfi")
|
|
(version "1.5.4")
|
|
(source
|
|
(origin
|
|
(method url-fetch)
|
|
(uri (string-append "https://gforge.inria.fr/frs/download.php"
|
|
"/latestfile/181/mpfi-" version ".tgz"))
|
|
(sha256
|
|
(base32 "0mismr1ll3wp788dq2n22s5irm0dziy75byyfdwz22kjbmckhf9v"))))
|
|
(build-system gnu-build-system)
|
|
(arguments
|
|
`(#:tests? #f ;tests are broken in this release
|
|
#:configure-flags '("--enable-static=no")))
|
|
(native-inputs
|
|
`(("automake" ,automake)
|
|
("autoreconf" ,autoconf)
|
|
("libtool" ,libtool)
|
|
("texinfo" ,texinfo)))
|
|
(propagated-inputs
|
|
(list gmp ; <mpfi.h> refers to both
|
|
mpfr))
|
|
(home-page "https://gforge.inria.fr/projects/mpfi/")
|
|
(synopsis "C library for arbitrary-precision interval arithmetic")
|
|
(description
|
|
"@acronym{MPFI, Multiple Precision Floating-point Interval} is a portable C
|
|
library for arbitrary-precision interval arithmetic, with intervals represented
|
|
using MPFR reliable floating-point numbers. It's based on the @acronym{GMP, GNU
|
|
Multiple Precision Arithmetic} and GNU@tie{}@acronym{MPFR, Multiple Precision
|
|
Floating-Point Reliably} libraries.
|
|
|
|
The purpose of arbitrary-precision interval arithmetic is to get results that
|
|
are both guaranteed, thanks to interval computation, and accurate, thanks to
|
|
multiple-precision arithmetic.")
|
|
(license lgpl2.1+)))
|
|
|
|
(define-public irram
|
|
(package
|
|
(name "irram")
|
|
(version "2013_01")
|
|
(source
|
|
(origin
|
|
(method url-fetch)
|
|
(uri (string-append "http://irram.uni-trier.de/irram-files/iRRAM_"
|
|
version ".tar.bz2"))
|
|
(sha256
|
|
(base32 "1cdmvb4hsa161rfdjqyhd9sb3fcr43p3a6nsj7cb4kn9f94qmjpj"))))
|
|
(build-system gnu-build-system)
|
|
(propagated-inputs (list gmp ; <mpfi.h> refers to both
|
|
mpfr))
|
|
(arguments
|
|
`(#:parallel-build? #f))
|
|
(synopsis "C++ package for real arithmetic based on the Real-RAM concept")
|
|
(description
|
|
"@dfn{iRRAM} is a C++ package for error-free real arithmetic based on
|
|
the concept of a Real-RAM. Its capabilities range from ordinary arithmetic
|
|
over trigonometric functions to linear algebra and differential
|
|
equations. A program using iRRAM is coded in ordinary C++, but may use a
|
|
special class that behaves like real numbers without any
|
|
error. Additionally, iRRAM uses the concept of multi-valued functions.")
|
|
(license lgpl2.0+)
|
|
(home-page "http://irram.uni-trier.de/")))
|
|
|
|
(define-public qd
|
|
(package
|
|
(name "qd")
|
|
(version "2.3.23")
|
|
(source
|
|
(origin
|
|
(method url-fetch)
|
|
(uri (string-append "https://crd-legacy.lbl.gov/~dhbailey/mpdist/qd-"
|
|
version ".tar.gz"))
|
|
(sha256
|
|
(base32 "09pfd77rmy370hy7qdqw84z21y9zpl3fcwzf93rhiv0kwhfg9smk"))))
|
|
(build-system gnu-build-system)
|
|
(native-inputs
|
|
(list gfortran))
|
|
(arguments
|
|
`(#:configure-flags `("--disable-enable_fma" ;weird :/
|
|
"--disable-static"
|
|
"--enable-shared"
|
|
,,@(if (string-prefix? "aarch64"
|
|
(or (%current-target-system)
|
|
(%current-system)))
|
|
;; XXX: The qd_test test fails numerical
|
|
;; accuracy checks for 'dd_real::exp()' on
|
|
;; aarch64 with GCC 5.4 at -O2. Disabling
|
|
;; expensive optimizations lets it pass.
|
|
'("CXXFLAGS=-O3 -fno-expensive-optimizations")
|
|
'("CXXFLAGS=-O3")))))
|
|
(home-page "https://www.davidhbailey.com/dhbsoftware/")
|
|
(synopsis "Double-double and quad-double library")
|
|
(description "This package supports both a double-double
|
|
datatype (approx. 32 decimal digits) and a quad-double datatype (approx. 64
|
|
decimal digits). The computational library is written in C++. Both C++ and
|
|
Fortran-90 high-level language interfaces are provided to permit one to
|
|
convert an existing C++ or Fortran-90 program to use the library with only
|
|
minor changes to the source code. In most cases only a few type statements
|
|
and (for Fortran-90 programs) read/write statements need to be changed. PSLQ
|
|
and numerical quadrature programs are included.")
|
|
(license bsd-3)))
|
|
|
|
(define-public cln
|
|
(package
|
|
(name "cln")
|
|
(version "1.3.6")
|
|
(source
|
|
(origin
|
|
(method url-fetch)
|
|
(uri (string-append "https://www.ginac.de/CLN/cln-"
|
|
version ".tar.bz2"))
|
|
(sha256
|
|
(base32 "0jlq9l4hphk7qqlgqj9ihjp4m3rwjbhk6q4v00lsbgbri07574pl"))))
|
|
(build-system gnu-build-system)
|
|
(arguments
|
|
`(#:configure-flags (list "--disable-static")))
|
|
(inputs
|
|
(list gmp))
|
|
(home-page "https://www.ginac.de/CLN/")
|
|
(synopsis "Library for arbitrary precision computations")
|
|
(description "CLN is a C++ library for efficient computations with all
|
|
kinds of numbers in arbitrary precision. It provides a rich set of number
|
|
classes and elementary, logical and transcendental functions.")
|
|
(license gpl2+)))
|
|
|
|
(define-public tomsfastmath
|
|
(package
|
|
(name "tomsfastmath")
|
|
(version "0.13.1")
|
|
(synopsis "Large integer arithmetic library")
|
|
(source (origin
|
|
(method url-fetch)
|
|
(uri (string-append "https://github.com/libtom/tomsfastmath/"
|
|
"releases/download/v" version "/"
|
|
"tfm-" version ".tar.xz"))
|
|
(sha256
|
|
(base32
|
|
"0f0pmiaskh89sp0q933pafxb914shpaj5ad8sb5rzk1wv8d7mja7"))))
|
|
(build-system gnu-build-system)
|
|
(native-inputs
|
|
(list libtool))
|
|
(arguments
|
|
`(#:make-flags (list "-f" "makefile.shared"
|
|
(string-append "LIBPATH=" %output "/lib")
|
|
(string-append "INCPATH=" %output "/include")
|
|
"GROUP=root" "USER=root"
|
|
"CC=gcc")
|
|
#:phases
|
|
(modify-phases %standard-phases
|
|
(delete 'configure) ; no configuration
|
|
(replace 'check
|
|
(lambda* (#:key make-flags #:allow-other-keys)
|
|
(apply invoke "make"
|
|
"stest" "test_standalone"
|
|
make-flags)
|
|
(invoke "./stest")
|
|
(invoke "./test")))
|
|
(add-before 'install 'install-nogroup
|
|
(lambda _
|
|
;; Let permissions inherit from the current process.
|
|
(substitute* "makefile.shared"
|
|
(("-g \\$\\(GROUP\\) -o \\$\\(USER\\)") ""))
|
|
#t))
|
|
(add-after 'install 'install-doc
|
|
(lambda* (#:key outputs #:allow-other-keys)
|
|
(let ((docdir (string-append (assoc-ref outputs "out")
|
|
"/share/doc/tomsfastmath")))
|
|
(install-file "doc/tfm.pdf" docdir)
|
|
#t)))
|
|
(add-after 'install 'install-pc
|
|
(lambda* (#:key outputs #:allow-other-keys)
|
|
(let* ((out (assoc-ref outputs "out"))
|
|
(pc-dir (string-append out "/lib/pkgconfig")))
|
|
(call-with-output-file "tomsfastmath.pc"
|
|
(lambda (port)
|
|
(format port "~
|
|
Name: TomsFastMath
|
|
Description: ~a
|
|
Version: ~a
|
|
Libs: -L~a/lib -ltfm~%"
|
|
,synopsis ,version out)))
|
|
(install-file "tomsfastmath.pc" pc-dir)
|
|
#t))))))
|
|
(home-page "https://www.libtom.net/TomsFastMath/")
|
|
(description "TomsFastMath is a large integer library written in portable
|
|
ISO C. It is a port of LibTomMath with optional support for inline assembler
|
|
multiplies.")
|
|
(license public-domain)))
|
|
|
|
(define-public libtomcrypt
|
|
(package
|
|
(name "libtomcrypt")
|
|
(version "1.18.2")
|
|
(outputs '("out" "static"))
|
|
(source
|
|
(origin
|
|
(method url-fetch)
|
|
(uri (string-append "https://github.com/libtom/libtomcrypt"
|
|
"/releases/download/v" version
|
|
"/crypt-" version ".tar.xz"))
|
|
(sha256
|
|
(base32
|
|
"113vfrgapyv72lalhd3nkw7jnks8az0gcb5wqn9hj19nhcxlrbcn"))
|
|
(modules '((guix build utils)))
|
|
(snippet
|
|
'(begin
|
|
;; Patch CVE-2019-17362
|
|
;; https://github.com/libtom/libtomcrypt/commit/25c26a3b7a9ad8192ccc923e15cf62bf0108ef94
|
|
(substitute* "src/pk/asn1/der/utf8/der_decode_utf8_string.c"
|
|
(("z > 4") "z == 1 || z > 4"))
|
|
#t))))
|
|
(build-system gnu-build-system)
|
|
(arguments
|
|
`(#:phases
|
|
(modify-phases %standard-phases
|
|
(delete 'configure) ; no configure
|
|
(add-after 'unpack 'prepare-build
|
|
(lambda _
|
|
;; We want the shared library by default so force it to be the
|
|
;; default makefile target.
|
|
(delete-file "makefile")
|
|
(symlink "makefile.shared" "makefile")
|
|
;; We link to libtommath, so we need to add it to the pc file
|
|
(substitute* "libtomcrypt.pc.in"
|
|
(("-ltomcrypt") "-ltomcrypt -ltommath"))
|
|
#t))
|
|
(add-after 'build 'build-static
|
|
(lambda* (#:key make-flags #:allow-other-keys)
|
|
(apply invoke "make" "-f" "makefile.unix" make-flags)))
|
|
(replace 'check
|
|
(lambda* (#:key test-target make-flags #:allow-other-keys)
|
|
(apply invoke "make" "-f" "makefile.unix" test-target make-flags)
|
|
(invoke "./test")))
|
|
(add-after 'install 'install-static-library
|
|
(lambda* (#:key outputs #:allow-other-keys)
|
|
(let ((out (assoc-ref outputs "out"))
|
|
(static (assoc-ref outputs "static")))
|
|
(mkdir-p (string-append static "/lib"))
|
|
(mkdir-p (string-append static "/include"))
|
|
(rename-file (string-append out "/lib/libtomcrypt.a")
|
|
(string-append static "/lib/libtomcrypt.a"))
|
|
(copy-recursively (string-append out "/include")
|
|
(string-append static "/include"))
|
|
#t))))
|
|
#:test-target "test"
|
|
#:make-flags
|
|
(list (string-append "PREFIX=" (assoc-ref %outputs "out"))
|
|
"CFLAGS += -DLTM_DESC -DUSE_LTM"
|
|
(string-append "EXTRALIBS="
|
|
(search-input-file %build-inputs
|
|
"/lib/libtommath.so"))
|
|
(string-append "CC=" ,(cc-for-target)))))
|
|
(native-inputs
|
|
(list libtool))
|
|
(inputs
|
|
(list libtommath))
|
|
(home-page "https://www.libtom.net/LibTomCrypt/")
|
|
(synopsis "Cryptographic toolkit")
|
|
(description "LibTomCrypt is a fairly comprehensive, modular and portable
|
|
cryptographic toolkit that provides developers with a vast array of well known
|
|
published block ciphers, one-way hash functions, chaining modes, pseudo-random
|
|
number generators, public key cryptography and a plethora of other routines.")
|
|
(properties `((lint-hidden-cve . ("CVE-2019-17362"))))
|
|
(license unlicense)))
|
|
|
|
(define-public libtommath
|
|
(package
|
|
(name "libtommath")
|
|
(version "1.2.0")
|
|
(outputs '("out" "static"))
|
|
(source
|
|
(origin
|
|
(method url-fetch)
|
|
(uri (string-append "https://github.com/libtom/libtommath/releases/"
|
|
"download/v" version "/ltm-" version ".tar.xz"))
|
|
(sha256
|
|
(base32
|
|
"1c8q1qy88cjhdjlk3g24mra94h34c1ldvkjz0n2988c0yvn5xixp"))))
|
|
(build-system gnu-build-system)
|
|
(arguments
|
|
'(#:phases
|
|
(modify-phases %standard-phases
|
|
(delete 'configure) ; no configure
|
|
(add-after 'unpack 'prepare-build
|
|
(lambda _
|
|
;; We want the shared library by default so force it to be the
|
|
;; default makefile target.
|
|
(delete-file "makefile")
|
|
(symlink "makefile.shared" "makefile")
|
|
#t))
|
|
(add-after 'install 'remove-static-library
|
|
(lambda* (#:key outputs #:allow-other-keys)
|
|
(delete-file (string-append (assoc-ref outputs "out")
|
|
"/lib/libtommath.a"))
|
|
#t))
|
|
(replace 'check
|
|
(lambda* (#:key test-target make-flags #:allow-other-keys)
|
|
(apply invoke "make" test-target make-flags)
|
|
(invoke "sh" "test")))
|
|
(add-after 'install 'install-static-library
|
|
(lambda* (#:key outputs #:allow-other-keys)
|
|
(invoke "make" "-f" "makefile.unix" "install"
|
|
(string-append "PREFIX=" (assoc-ref outputs "static"))
|
|
(string-append "CC=" (which "gcc"))))))
|
|
#:test-target "test"
|
|
#:make-flags (list (string-append "PREFIX=" (assoc-ref %outputs "out"))
|
|
"CC=gcc")))
|
|
(native-inputs
|
|
(list libtool))
|
|
(home-page "https://www.libtom.net/LibTomMath/")
|
|
(synopsis "Portable number theoretic multiple-precision integer library")
|
|
(description "LibTomMath is a portable number theoretic multiple-precision
|
|
integer library written entirely in C. It's designed to provide an API that is
|
|
simple to work with that provides fairly efficient routines that build out of
|
|
the box without configuration.")
|
|
(license unlicense)))
|
|
|
|
(define-public libtommath-1.1
|
|
(package
|
|
(inherit libtommath)
|
|
(version "1.1.0")
|
|
(source
|
|
(origin
|
|
(method url-fetch)
|
|
(uri (string-append "https://github.com/libtom/libtommath/releases/"
|
|
"download/v" version "/ltm-" version ".tar.xz"))
|
|
(sha256
|
|
(base32
|
|
"1bbyagqzfdbg37k1n08nsqzdf44z8zsnjjinqbsyj7rxg246qilh"))
|
|
(patches (search-patches "libtommath-fix-linkage.patch"))))
|
|
(arguments
|
|
(substitute-keyword-arguments (package-arguments libtommath)
|
|
((#:phases phases)
|
|
`(modify-phases ,phases
|
|
(add-after 'unpack 'patch-coreutils-call
|
|
(lambda _
|
|
;; Don't pull in coreutils.
|
|
(substitute* "makefile_include.mk"
|
|
(("arch") "uname -m"))
|
|
#t))))
|
|
((#:test-target _) "test_standalone")))))
|
|
|
|
(define-public libtommath-1.0
|
|
(package
|
|
(inherit libtommath-1.1)
|
|
(version "1.0.1")
|
|
(outputs '("out"))
|
|
(source
|
|
(origin
|
|
(method url-fetch)
|
|
(uri (string-append "https://github.com/libtom/libtommath/releases/"
|
|
"download/v" version "/ltm-" version ".tar.xz"))
|
|
(sha256
|
|
(base32
|
|
"0sbccdwbkfc680id2fi0x067j23biqcjqilwkk7y9339knrjy0s7"))))
|
|
(arguments
|
|
(substitute-keyword-arguments (package-arguments libtommath-1.1)
|
|
((#:phases phases)
|
|
`(modify-phases ,phases
|
|
(delete 'install-static-library)))))))
|