Takaya Uchida

I personally support such functionality (from selfish motives).

This is indeed a cool idea but I have no idea how much extra work this entails...

Thanks for using our package @iuryt :) In Appendix E of this paper, we provide how to estimate the 95% confidence level based on the chi2 distribution: https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1029/2021MS002692 Jupyter notebooks...

Hi @santisoler, thanks for your interest in `xrft` and detailed feedback. We are happy to know that there are people using our package. > ## Differences between `dft` and `fft`...

> I understand. My question is: does it worth having two different functions that are intended to do the same thing with slight differences? > > IMHO, I find it...

> IMO we should strongly consider deprecating one or the other. `dft` is not a standard function name, while `fft` appears in many different packages. I think the first step...

> > IMO we should strongly consider deprecating one or the other. `dft` is not a standard function name, while `fft` appears in many different packages. I think the first...

> > `dft` is the legacy naming in xrft so if we are to drop either `fft` or `dft`, for compatibility with prior versions, I vote for dropping `fft`. >...

> I like the round-tripping argument :) > @roxyboy, I can start this PR too if you need Sure, if you could start a PR about keeping the original coordinates...

Perhaps this has to do with xrft having trouble computing the frequency coordinate when the original data has no coordinate metadata?