Evgeni Burovski

Note to self: tutorial files are not only in `doc/source/tutorial`, also `doc/source/tutorial/stats` etc, so need to recurse into subdirs.

@lucascolley mostly superseded, yes. I plan to take a closer look at gh-19242 in the next couple of weeks. Meanwhile, feel free to close this one if it gets in...

Definitely a big +1 from me. What we have currently is historical, and a robust method from Trefethen and co is very much welcome as a replacement. I think it's...

cross-ref https://github.com/scipy/scipy/pull/21167 --- @j-bowhay is the AAA PR what you had in mind here (so that it closes this one) or do you think it'd make sense to improve `pade`...

Bumped into this again in #19970. In the 1.14 timeframe this was enough ``` #include "../_build_utils/src/fortran_defs.h" #define DLARTG F_FUNC(dlartg, DLARTG) ``` but not anymore.

API are frozen but ABI aren't, and that's I suppose just a fact of life. It's just that something has changed recently, either in scipy or in the GNU compiler...

> ABI stability should be present since this is a pure C interface without anything mangled Oh there definitely is some name mangling. At least `scipy_openblas32` renames things to have...

> Writing interfaces for all of LAPACK and BLAS is a daunting task, but there was a GSoC project some years ago trying to automate it. I'm happy to test-drive...

Where does this PR stand, now that 1.14 is closing in?

This line has been there forever (tm): ``` $ git blame -M -C -C scipy/interpolate/fitpack/pardeu.f eda24252c92 Lib/interpolate/fitpack/parder.f (Travis Oliphant 2001-06-20 06:23:55 +0000 82) if(nux.lt.0 .or. nux.ge.kx) go to 400 ```...