Shouldn't bzr_src_unpack return to ${WORKDIR}{$P} after it's done unpacking? An eautoreconf fails after unpacking with the following error unless I perform a cd ${WORKDIR}/${P} after the bzr_src_unpack command. * Failed Running aclocal ! * * Include in your bugreport the contents of: * * /var/tmp/portage/sys-fs/ltspfs-9999/temp/aclocal.out the aclocal.out: ***** aclocal ***** ***** PWD: /var/tmp/portage/sys-fs/ltspfs-9999/work ***** aclocal aclocal-1.10: `configure.ac' or `configure.in' is required
That's ${S}, but yes, I think. Anyone has another opinion? Though I cannot see why it should not be in S after calling the function as it is a local directory change and we use popd to get back again to the inital directory.
(In reply to comment #1) > That's ${S}, but yes, I think. Anyone has another opinion? Though I cannot see > why it should not be in S after calling the function as it is a local directory > change and we use popd to get back again to the inital directory. Christian, I agree we should cd to ${S}. When src_unpack is called, there's nothing under ${WORKDIR}, so the final popd gets us back to ${WORKDIR}. However, shouldn't src_prepare start in ${S}? As bzr_bootstrap starts by pushing ${S} and finishes by poping out of it, if the eautoreconf is being called after the call to bzr_src_prepare, the current dir will be the dir src_prepare starts in. Wim, can we see the ebuild you're using? What EAPI are you using and are you using the default bzr_src_prepare or are you overriding it?
sys-fs-ltspfs/ltspfs-9999 from the LTSP overlay. Don't mind the comments, its still in development. no EAPI specified, bzr_src_prepare is called from bzr_src_unpack # Copyright 1999-2008 Gentoo Foundation # Distributed under the terms of the GNU General Public License v2 # $Header: $ if [[ ${PV} == 9999* ]] then [[ ${PV} == 9999.* ]] && EBZR_REVISION="${PV/9999./}" EBZR_REPO_URI="http://bazaar.launchpad.net/~ltsp-upstream/ltspfs/ltspfs-trunk" inherit bzr autotools else inherit autotools SRC_URI="mirror://gentoo/${P}.tar.bz2" fi DESCRIPTION="LTSP file system" HOMEPAGE="http://www.ltsp.org/" LICENSE="GPL-2" SLOT="0" KEYWORDS="~x86 ~amd64" IUSE="" DEPEND=">=dev-libs/glib-2.6" RDEPEND="${DEPEND} >=sys-fs/fuse-2.7.2 x11-libs/libX11" src_unpack() { if [[ ${PV} == 9999* ]] ; then bzr_src_unpack cd ${WORKDIR}/${P} # FIXME: do we want to use bzr to generate a ChangeLog? # Create empty ChangeLog to please autoreconf touch ChangeLog else unpack ${P}.tar.bz2 fi eautoreconf } src_install() { emake DESTDIR="${D}" install || die "einstall failed" # TODO: make NEWS and README install dodoc NEWS README docinto examples dodoc doc/examples/* }
(In reply to comment #3) > sys-fs-ltspfs/ltspfs-9999 from the LTSP overlay. Don't mind the comments, its > still in development. > > no EAPI specified, bzr_src_prepare is called from bzr_src_unpack > > if [[ ${PV} == 9999* ]] > then > [[ ${PV} == 9999.* ]] && EBZR_REVISION="${PV/9999./}" PV is a read-only var. You'll want to use MY_PV instead. What are you trying to do with the replacement? It doesn't seem correct. > EBZR_REPO_URI="http://bazaar.launchpad.net/~ltsp-upstream/ltspfs/ltspfs-trunk" s/ltspfs/${PN}/ > KEYWORDS="~x86 ~amd64" It follows alphabetical order so KEYWORDS="~amd64 ~x86" and you should only add keywords for arches you test. > src_unpack() { > if [[ ${PV} == 9999* ]] ; then > bzr_src_unpack > cd ${WORKDIR}/${P} > > # FIXME: do we want to use bzr to generate a ChangeLog? > # Create empty ChangeLog to please autoreconf > touch ChangeLog > else > unpack ${P}.tar.bz2 > fi If you rely on the default src_unpack you can avoid src_unpack altogether as bzr_unpack will be called automatically if you inherit bzr. > eautoreconf This should be called from src_prepare. Given your current ebuild, going straight for EAPI-2 seems to make more sense. You may want to look at http://devmanual.gentoo.org/ for more info about ebuilds.
The src_prepare and the EAPI 2 did the trick. It works now. I'll also look into the other problems. Thanks for the support.
