Evgeni Burovski

Overall, this sounds reasonable. A big question is, like Pamphile said, how to do it in a clean way with pythran implementations. A pull request is most welcome @smeingast !...

Rbf specifically we can deprecate I think if there's a general preference. While I'm not an rbf user, I suspect that RBFInterpolator offers a strict superset, fixes long-standing Rbf problems...

FWIW, I'd think it's reasonable ask to postpone the deprecation until the next release cycle, so that Rbf starts emitting deprecation warnings around summer 2023. Would you please respond to...

Re notebook bitrot, MinRK's script seems useful (and can be extended with our doctest checker from refguide_check.py if desired). http://stackoverflow.com/questions/20483313/testing-ipython-notebooks

My slight preference is still for a separate repo: - bitrotting is an issue regardless of where things are - incorporating notebooks into the build sounds fragile. Josef's comment "nbconvert...

As a drive-by comment, when there's a comparison to other tools, it'd be nice to compare to `approx_derivative`. (I realize these two are very different)

>Polynomial extrapolation is generalized Richardson extrapolation. Richardson is linear extrapolation. Polynomial extrapolation uses a higher order polynomial. How do you select the defeee of this polynomial? What controls the extrapolation...

@axil nice write-up! Am planning to link to it from the scipy user guide then in gh-16707 or a follow-up. Are you're open to amending/expanding your write-up? (reopening for visibility)

@tylerjereddy @axil Too bad I did not make it to the SciPy meeting (was 11pm local time for me). Doc improvements are definitely *very* welcome! A kwarg is very much...

@axil this nearly slipped under the radar, sorry about that. Your text is great actually. A couple of minor comments first: 1D: interp1d does not use Fortran. It's pure python...