search

ian-ross / arb-fft

Pure Haskell arbitrary length FFT library
https://www.skybluetrades.net/haskell-fft-index.html
BSD 3-Clause "New" or "Revised" License
11 stars 2 forks source link

Bring up to date with ghcjs-0.2 and criterion 1.1 #7

Closed imalsogreg closed 8 years ago

imalsogreg commented 8 years ago

These changes allow arb-fft to compile and tests to pass under ghc 7.10.2 and all the dependency constraints coming along with ghcjs.

The changes to Numeric.FFT.Plan might change the way criterion operates - criterion >= 1.0 dropped some types like Environment without a clear replacement, so the caching of environment measurement became impossible, and that measurement code was removed. Is this Ok?

ian-ross commented 8 years ago

@imalsogreg Thanks for this! I'll take a look at it in the next few days. I need to figure out why the Travis tests are failing and think about the Criterion issue you pointed out.

ian-ross commented 8 years ago

@imalsogreg Looks good! Not sure why the Travis tests were failing. They work locally, so it may just be a Travis environment issue. And the changes to Criterion look perfectly fine. Thanks for the PR!

ian-ross commented 8 years ago

@imalsogreg Actually, there was one thing I needed to fix, which was the thing that was causing the test failures on Travis. The tests were timing out because planning was taking so long (you don't notice running the tests locally because you usually have plan wisdom files lying around in ~/.fft-plan). This was down to changes in the way that Criterion controls the number of iterations it uses for benchmarking. I've implemented a slightly low-rent solution, but it seems to work OK.

imalsogreg commented 8 years ago

Excellent - thanks a lot for the investigation and the fix!