fftw ebuild for 3.0, using slot '3.0' and no MPI

fftw-3.0.ebuild (text/plain), 3.75 KB, created by Ingo Luetkebohle on 2003-06-18 14:21:30 UTC

# Copyright 1999-2003 Gentoo Technologies, Inc.
# Distributed under the terms of the GNU General Public License v2
# $Header: /home/cvsroot/gentoo-x86/dev-libs/fftw/fftw-2.1.5.ebuild,v 1.2 2003/05/07 22:47:54 george Exp $

inherit flag-o-matic

S="${WORKDIR}/${P}"
DESCRIPTION="C subroutine library for computing the Discrete Fourier Transform (DFT)"
SRC_URI="http://www.fftw.org/${P}.tar.gz"
HOMEPAGE="http://www.fftw.org"

SLOT="3.0"
LICENSE="GPL-2"

#remove ~'s on ppc and sparc when removig on x86 (as per recent discussion on -core)
KEYWORDS="x86 ppc sparc alpha"

#this one is reported to cause trouble on pentium4 m series
filter-flags "-mfpmath=sse"

#here I need (surprise) to increase optimization:
#--enable-i386-hacks requires -fomit-frame-pointer to work properly
export CFLAGS="${CFLAGS/-fomit-frame-pointer/} -fomit-frame-pointer"

pkg_setup() {
	einfo ""
	einfo "This ebuild installs double and single precision versions of library"
	einfo "This involves some name mangling, as supported by package and required"
	einfo "by some apps that use it."
	einfo "By default, the symlinks to non-mangled names will be created off"
	einfo "double-precision version. In order to symlink to single-precision use"
	einfo "SINGLE=yes emerge fftw"
	einfo ""
}

src_unpack() {
	#doc suggests installing single and double precision versions via separate compilations
	#will do in two separate source trees
	#since some sed'ing is done during the build (?if --enable-type-prefix is set?)

	unpack "${P}.tar.gz"
	cd "${WORKDIR}"
	mv ${P} ${P}-single

	unpack "${P}.tar.gz"
	cd "${WORKDIR}"
	mv ${P} ${P}-double
}


src_compile() {
	local myconf=""
	use mpi && myconf="${myconf} --enable-mpi"
	#mpi is not a valid flag yet. In this revision it is used merely to block --enable-mpi option
	#it might be needed if it is decided that lam is an optional dependence

	cd "${S}-single"
	econf \
		--enable-shared \
		--enable-threads \
		--enable-type-prefix \
		--enable-float \
		--enable-i386-hacks \
		--enable-vec-recurse \
		${myconf} || die "./configure failed"
	emake || die

	#the only difference here is no --enable-float
	cd "${S}-double"
	econf \
		--enable-shared \
		--enable-threads \
		--enable-type-prefix \
		--enable-i386-hacks \
		--enable-vec-recurse \
		${myconf} || die "./configure failed"
	emake || die
}

src_install () {
	#both builds are installed in the same place
	#libs are distinguished by preffix (s or d), see docs for details
	cd "${S}-single"

	make DESTDIR=${D} install || die

	cd "${S}-double"

	# fix info file
	echo "INFO-DIR-SECTION Libraries" >>$fftw.info
	echo "START-INFO-DIR-ENTRY" >>doc/fftw.info
	echo "* fftw: (fftw).                  C subroutine library for computing the Discrete Fourier Transform (DFT)" >>doc/fftw.info
	echo "END-INFO-DIR-ENTRY" >>doc/fftw.info

	make DESTDIR=${D} install || die

	# Install documentation.
	cd "${S}-single"

	dodoc AUTHORS ChangeLog COPYING INSTALL NEWS  TODO
	dohtml doc/fftw*.html

	if [ "$SINGLE" = "yes" ]; then
		cd ${D}usr/include
		dosym sfftw.h /usr/include/fftw.h
		dosym srfftw.h /usr/include/rfftw.h
		dosym sfftw_threads.h /usr/include/fftw_threads.h
		dosym srfftw_threads.h /usr/include/rfftw_threads.h
		cd ${D}usr/lib
		dosym libsfftw.so /usr/lib/libfftw.so
		dosym libsrfftw.so /usr/lib/librfftw.so
		dosym libsfftw_threads.so /usr/lib/libfftw_threads.so
		dosym libsrfftw_threads.so /usr/lib/librfftw_threads.so
	else
		cd ${D}usr/include
		dosym dfftw.h /usr/include/fftw.h
		dosym drfftw.h /usr/include/rfftw.h
		dosym dfftw_threads.h /usr/include/fftw_threads.h
		dosym drfftw_threads.h /usr/include/rfftw_threads.h
		cd ${D}usr/lib
		dosym libdfftw.so /usr/lib/libfftw.so
		dosym libdrfftw.so /usr/lib/librfftw.so
		dosym libdfftw_threads.so /usr/lib/libfftw_threads.so
		dosym libdrfftw_threads.so /usr/lib/librfftw_threads.so
	fi
}