As the title says, version bump for this. Finally it gets rid of the old numeric dependency, and relies only on numpy. This also fixes the compilation problems in the kdtree module as far as I have read in the mailing lists and that, so I removed it as well. Ebuild and patch is coming. Reproducible: Always
Created attachment 173372 [details] sci-biology/biopython-1.49.ebuild This is the ebuild. Apart for the kdtree stuff, I added optional dependencies on mysql and postgres modules, to use the app with BioSQL databases. However, the test on those parts fails because it assumes that a BioSQL database exists, also the test file must be changed for each user with the login, password, etc. for that DB, so I leave it as is. reportlab and flex seems to be optional dependencies, but the repotlab stuff seems nearly essential (according to ML, docs and the like) and the flex one only depends on that, and is optional because it doesn't compiles on windows. The test for the module that depends on reportlab, GFF, fails when being tested on reportlab-1*, but pass as GFF2 when using reportlab-2*. I don't know if this is something expected, or if you need both versions of reportlab, or what, but I think that this is OK. Note that the test is always being run with this ebuild. I don't know how to handle testing in the ebuilds, I've tried using the "test" useflag but it didn't work. Please fix it if needed.
Created attachment 173373 [details, diff] biopython-1.49-flex.patch This is the patch needed for the ebuild. It makes the program able to install the module that depends on flex, it was commented because it doesn't compile on windows or cygwin. Note that I don't use this app and I don't know how I can test this, so I'm trusting the provided test suite. I'm just want to get rid of numeric and numarray :)
Thanks for your contributions! ribosome, will you be able to take care of this or should I take a look at bumping it?
(In reply to comment #3) > Thanks for your contributions! ribosome, will you be able to take care of this > or should I take a look at bumping it? > I'll look at this during the holidays, but feel free to beat me to it. :) Olivier
I see during testing of this ebuild: >>> Completed installing biopython-1.49 into /var/tmp/portage/sci-biology/biopython-1.49/image/ strip: i686-pc-linux-gnu-strip --strip-unneeded -R .comment usr/lib/python2.5/site-packages/Bio/cmathfns.so usr/lib/python2.5/site-packages/Bio/cstringfns.so usr/lib/python2.5/site-packages/Bio/Cluster/cluster.so usr/lib/python2.5/site-packages/Bio/cMarkovModel.so usr/lib/python2.5/site-packages/Bio/clistfns.so usr/lib/python2.5/site-packages/Bio/cpairwise2.so usr/lib/python2.5/site-packages/Bio/Restriction/DNAUtils.so usr/lib/python2.5/site-packages/Bio/trie.so usr/lib/python2.5/site-packages/Bio/PDB/mmCIF/MMCIFlex.so usr/lib/python2.5/site-packages/Bio/Nexus/cnexus.so usr/lib/python2.5/site-packages/Bio/KDTree/_CKDTree.so * QA Notice: Precompiled python object files do not belong in /usr/share * /usr/share/biopython/Tests/test_Cluster.pyc * /usr/share/biopython/Tests/test_NCBIXML.pyc * /usr/share/biopython/Tests/test_HMMGeneral.pyc * /usr/share/biopython/Tests/test_KEGG.pyc * /usr/share/biopython/Tests/test_PDB.pyc * /usr/share/biopython/Tests/test_BioSQL.pyc [cut] I also think it is a waste of space to install pre-compile testing files.
In Portage. Thanks for your work, Juan, and thanks for the testing, Martin. I fixed the *.pyc issue and removed FEATURES=test, as ebuilds should not set FEATURES. The test USE flag should not be used either. Users can enable src_test with FEATURES=test in their make.conf.
Please stabilize sci-biology/biopython-1.49 for x86 and amd64. almost 5 months in the tree.
