New release requires no real changes. I do however have a much better patch for the Autoconf 2.6[23] problem, which has been accepted upstream for the same bug in libsndfile, so there is a new patch to go with the ebuild patch.
Created attachment 182270 [details, diff] Patch to the 0.1.6 ebuild for 0.1.7 Very simple ebuild patch.
Created attachment 182271 [details, diff] Patch to fix autoconf issues This is a single [ move, but without it we get Autoconf failures. It is superior to the sed invocation because it doesn't break the upstream functionality.
thanks; everything seems good but I get a (stupid?) test failure: Sinc interpolator : callback_test (SRC ratio = 1.0000) ........... ok callback_test (SRC ratio = 0.0990) ........... ok callback_test (SRC ratio = 0.1000) ........... ok callback_test (SRC ratio = 0.3333) ........... ok callback_test (SRC ratio = 0.7890) ........... Line 141 : input / output length mismatch. input len : 10000 output len : 7889 (should be 7890 +/- 1)
Test fails on a 64-bit platform? This has been reported on the development list, where the response from Erik was "Yeah, its some weird floating point rounding issue. It can safely be ignored. I'll get a fix out for it.". Disable tests in the ebuild in the mean time? We could patch the test suite to remove that test, but I don't think it's worth the effort.
(In reply to comment #4) > Test fails on a 64-bit platform? Yes exactly > This has been reported on the development > list, where the response from Erik was "Yeah, its some weird floating point > rounding issue. It can safely be ignored. I'll get a fix out for it.". Bah... I'm now subscribed to the ml but couldn't find any archive; is there a patch available yet? > Disable > tests in the ebuild in the mean time? I do like tests :/ I would really prefer seeing this fixed. > We could patch the test suite to remove > that test, but I don't think it's worth the effort. Or wait for Erik's fix and apply it; to me this seems the simplest idea. Of course if it takes too long the test can be removed but that'd be cheating :)
> > This has been reported on the development > > list, where the response from Erik was "Yeah, its some weird floating point > > rounding issue. It can safely be ignored. I'll get a fix out for it.". > > Bah... I'm now subscribed to the ml but couldn't find any archive; is there a > patch available yet? There isn't a patch, and unfortunately I don't know of an archive yet (I only saw the report because it was the first message that came down the list when I joined). > Or wait for Erik's fix and apply it; to me this seems the simplest idea. Of > course if it takes too long the test can be removed but that'd be cheating :) I have no feel yet for what the timescale for a patch might be. Because there isn't a public VCS it's quite likely that the first fix we can easily get hold of is the release of a 0.1.8 pre-release as a tarball, which would have to be diff'ed to get a patch.
bumped thanks; I've added a patch from Erik to fix the tests.
