PAML is an excellent set of programs for phylogenetic analyses all written in good ol' C. It can make many kinds of analyses using the Maximum Likelihood approach (the cool wave among phylogeneticists), and is the best one I know on this field. Of course I suggest app-sci (isn't this redundant?). This package apparently has no dependencies. As this is my first ebuild ever, excuse me if something is terrible. I read through all the necessary docs, but I still am not so sure about this ebuild's correctness. Reproducible: Always Steps to Reproduce: 1. 2. 3.
Created attachment 15515 [details] paml-3.13d.ebuild (New Package)
Comment on attachment 15515 [details] paml-3.13d.ebuild (New Package) # Copyright 1999-2003 Gentoo Technologies, Inc. # Distributed under the terms of the GNU General Public License v2 DESCRIPTION="PAML (Phylogenetic Analyses by Maximum Likelihood) is a package of programs for analyzing and inferring phylogenetic data" HOMEPAGE="http://abacus.gene.ucl.ac.uk/software/paml.html" SRC_URI="ftp://abacus.gene.ucl.ac.uk/pub/paml/paml3.13d.tar.gz" LICENSE="GPL" SLOT="0" KEYWORDS="x86" IUSE="" DEPEND="" src_compile() { cd ${WORKDIR}/paml3.13d/src cp Makefile.UNIX Makefile # uses only makefile. # trying to use make.conf's CFLAGS EXTRA_EMAKE="-e ${CFLAGS}" emake || die } src_install() { cd ${WORKDIR}/paml3.13d/src DESTTREE=/usr/local/${P} dobin baseml basemlg codeml pamp evolver yn00 chi2 cd doc dodoc pamlDOC.pdf pamlHistory.txt paml.html } pkg_postinst() { cp ${WORKDIR}/paml3.13d/* --recursive /usr/local/${P} mv /usr/local/${P}/bin/* /usr/local/${P} rm -rf /usr/local/${P}/{src,bin} rm /usr/local/${P}/*.exe }
Created attachment 15516 [details] I corrected the header, which I had previously copied from another ebuild file.
Sorry for this whole mess. Ebuild newbie here. Please ignore "Additional Comment #2". I feel so stupid.
Please change your ebuild - nothing should go into /usr/local since all ebuilds provide parts of the distribution. I am rather sure that you don't really need post installation steps, what you're trying there looks more like trying to clean up what went wrong during install. You should also set ${S} instead of cd-ing there every time.
I got a huge problem with this program, and I think I'll never make it work, unless I change its code. The binaries need control files (for example, baseml needs baseml.ctl) to be located on the same directory as the binaries themselves. This would be quite difficult to achieve, since binaries go to /usr/bin. The thing is, people who use this software always run it from their home directories. Also, the whole software package is pretty easy to compile. So, I think there's no actual _need_ for this package to be on Portage, besides my obvious inaptitude to change its code. I quit. If anyone feels like hacking it, I'll be glad to help as possible.
