Gaurav Arya

I could revert to the submodule -- my worry there is that if we come up with a better solution in the future, we couldn't remove the submodule from `AbstractFFTs`...

Also, I just want to make sure I understand the issue you pointed out with a separate package. If I'm understanding right, it would occur if a breaking change was...

@sethaxen I saw that you've been working on test utilities for Enzyme (https://github.com/EnzymeAD/Enzyme.jl/pull/782), so I was wondering if you had input on the best design here for `AbstractFFTsTestUtils`, between a...

@devmotion @sethaxen This is now ready! The test suite includes tests of adjoint functionality which would hopefully be helpful downstream (see https://github.com/JuliaMath/AbstractFFTs.jl/issues/94, I've tested that the test suite also works...

Comments addressed, and I also refactored some repeated logic in the test utilities 🙂

Bump on review

@vpuri3 confused me at first too, but that is how to migrate from a hard dep to a weak extension on 1.9+: https://pkgdocs.julialang.org/v1.9/creating-packages/#Transition-from-normal-dependency-to-extension

does this PR need anything more to merge? Please note that I have locally tested https://github.com/JuliaMath/FFTW.jl/pull/249 with https://github.com/JuliaMath/AbstractFFTs.jl/pull/78 and https://github.com/JuliaMath/AbstractFFTs.jl/pull/109, so FFTW rules are ready to go. See also the...

@ZuseZ4 tells me this is innocuous (only ran into this after significant minimization)

Here is a full application where we still run into the same error. ```C // this MUST be first, otherwise there might be problems on windows // see: https://stackoverflow.com/questions/6563810/m-pi-works-with-math-h-but-not-with-cmath-in-visual-studio/6563891#6563891 #define...