Numerical uncertainty causing Travis CI to fail about appleaccelerate.jl

[simonbyrne]

Thanks for the interest!

So, the problem is a little more complicated than floating point error: the line that's failing on Travis is actually this one, the purpose of which was to check that we're actually calling the correct method after overwriting the Base.sin method. Unfortunately, it seems like this isn't actually working correctly.

More generally, these tests should probably be rewritten to use BaseTextNext.jl, as FactCheck is slowly being deprecated.

[rprechelt]

Ah; that makes sense! This is what happens when I just glance over this stuff late at night! My apologies for the "false" issue notice.

That being said, I'd be happy to tackle rewriting these tests using BaseTextNext.jl; I have been using AppleAccelerate.jl a lot for the last few weeks and would love to give back.

I have already implemented a number of other Accelerate functions (conv, xcorr, acorr and others) in my local repo; if these are functions you think we should merge in, I'll write some tests for them and make a PR.

[simonbyrne]

That would be great, thanks, contributions are certainly welcome. I think it would make sense to expose as much of Accelerate as possible.