I've committed gmpy-1.03.ebuild to the science overlay. The package has a new homepage, and its building is now more standard. I added src_test(). Not all tests pass on my computer. python (running tests) always returns 0, so, adding || die "tests failed" is meaningless. I found that running the tests-driving program gmpy_test.py produces more errors that running each gmpy_test_*.py separately. Maybe, something's wrong with the driving program, or tests interfere with each other. By running gmpy_test_*.py separately, I found errors only in gmpy_test_mpf.py. The first of them: >>>_g.mpf(3.4) mpf('3.39999999999999991118e0') expected: mpf('3.4e0') and many other similar errors. I don't know wether they should be taken seriously or not. In any case, it seems worthwhile to add gmpy-1.03 to the main tree. If you agree, I can do this myself.
please attach your ebuild here. I'll check it. thanks!
Created attachment 164436 [details] gmpy-1.03.ebuild Here it is
new version in CVS.
I didn't used your ebuild though, because it wasn't complete (just compare the 1.01 ebuild with yours). Furthermore: I hate failing tests because of crappy tests. That will give us just evil bug reports from our users. ;-)
