On Wed, 5 Feb 2020 at 17:40, deb via Boost
I was myself not quite sure of the idea of having a textbook implementation of FFT being useful for Boost.
As I explained, a classic implementation of FFT will be useful, but it is not yet really. First, we need to get (discuss, design and implement) the FFT adaptation support. If you are dying hard to submit such FFT to GIL soon, then it could be accepted as as an example (.hpp & .cpp files similar to the interleaved_ptr.hpp|cpp) or as an extension in boost/gil/extension (with tests and docs!). That, however, still does not qualify for a GSoC project.
I will right away pick up a simpler topic from the project proposal page on Github (probably smoothing/blurring) and spend time getting more familiar with the Boost.gil API. And as suggested by Mateusz Loskot , I will also try to add docs and tests to the implementation before I submit a pull request.
Sounds good, just please keep in mind that for competency we do not ask you to deliver fully-featured, tested and documented implementation of an algorithm. Leave some work for the actual GSoC :-) Best regards, -- Mateusz Loskot, http://mateusz.loskot.net