New biopython has improved the parsers in terms of ability to parse some features in various fileformats as well as speed gain and memory requirements. Just copy the 1.42 ebuild under 1.43 and apply this patch to setup.py to build KDTree.
Created attachment 119699 [details, diff] patch-to-build-KDTree.patch
Created attachment 119701 [details, diff] patch-to-build-KDTree.patch sorry, now it gets picked up by epatch()
Created attachment 119703 [details] biopython-1.43.ebuild working ebuild on ~x86
*biopython-1.43 (17 May 2007) 17 May 2007; Olivier Fisette <ribosome@gentoo.org> +files/biopython-1.43-buildkdtree.patch, -biopython-1.40b.ebuild, +biopython-1.43.ebuild: New upstream version. Ebuild clean-up. Pruned old version. Anything else you want to be done? `if 1 == 1` doesn't look like acceptable patch honestly. :)
OK, the official ebuild for 1.43 does not build the KDTree module for me. The patch only makes the build process to be non-interactive. On linux it builds fine for me so I do not see a reason not to build it. BTW, somebody on the biopython list recently reported that it is required by ... some package. ;-)
The ebuild in Portage works fine for me, KDTree support being controled by the kdtree USE flag. I decided not to enable it by default since upstream says it does not compile cleanly on all arches, but I could change this default later on (or even remove the flag) if it seems to work fine on all arches where we support Biopython. If you have additional information about KDTree on Alpha, PPC, Sparc or Linux in general, send it my way. If the current ebuild really seems broken for you, feel free to reopen and post the output of "emerge --info". Thanks,
