Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

WIP: Support 0.6 #13

Merged
merged 7 commits into from
Feb 3, 2017
Merged

WIP: Support 0.6 #13

merged 7 commits into from
Feb 3, 2017

Conversation

dlfivefifty
Copy link
Member

This starts supporting 0.6. I've dropped 0.4 compatibility so that the f.(x) syntax can be used. Are we OK with this change? (I think @compat can restore 0.4 compatibility but I'm not sure it's worth it.)

the BigFloat FFT test fails for some reason:

Testing BigFloat support of FFT and DCT methods
Test Failed
  Expression: norm(ifft(fft(c)) - c) < 200 * (norm(c) * eps(BigFloat))
   Evaluated: 3.290907383297889365567835493403282408529556851637210612211926363351574968498589e-13 < 2.841706296661406868731913357740366307573017126122182078005142968265071086475453e-74
ERROR: LoadError: LoadError: There was an error during testing
while loading /Users/solver/.julia/v0.6/FastTransforms/test/fftBigFloattest.jl, in expression starting on line 9
while loading /Users/solver/.julia/v0.6/FastTransforms/test/runtests.jl, in expression starting on line 189

• Drops 0.4 support
@dlfivefifty
Copy link
Member Author

We can't merge this until we have a solution to JuliaLang/Compat.jl#312

@dlfivefifty
Copy link
Member Author

As 0.5 compatibility is broken

@stevengj
Copy link

stevengj commented Feb 1, 2017

I'm confused by your bigfloat FFT test. Julia has never included a bigfloat FFT (if you pass it bigfloats, it converts to Float64). I had an rough PR for this at one point, but it was never merged.

@dlfivefifty
Copy link
Member Author

Mikael implemented an FFT for BigFloats: FastTransforms/src/fftBigFloat.jl

@dlfivefifty
Copy link
Member Author

dlfivefifty commented Feb 1, 2017

The implementation would also work for other types... @stevengj do you think it would be worth creating an FFT.jl package and putting it there (until a more robust implementation is developed)?

@stevengj
Copy link

stevengj commented Feb 1, 2017

See also JuliaLang/julia#6193 for my bigfloat implementation, which also used Bluestein combined with radix-2 IIRC ... yes, but at some point the fft functionality is likely to be moved out of Base, so maybe that would be a better time to make an FFT package.

@MikaelSlevinsky
Copy link
Member

I don't have an issue with this. The BigFloat FFT hasn't been optimized, but it has been useful from time to time.

@dlfivefifty
Copy link
Member Author

This is actually a bug in Julia 0.6: JuliaLang/julia#20381

I can't work around it using colon: JuliaLang/julia#20382

@dlfivefifty
Copy link
Member Author

The bug was fixed in 0.6 master

@dlfivefifty dlfivefifty merged commit 6081670 into master Feb 3, 2017
@dlfivefifty dlfivefifty deleted the support-0.6 branch February 3, 2017 10:23
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants