Albert Steppi

> > > CI failure looks real > > > > > > Interesting. It looks like that job was added only a few weeks ago. I'll have to look...

Ok, so there were two test failures. One was `test_cython_special.py` timing out. This actually had nothing to do with `cython_special`, and was actually because `wright_bessel(10.0, 1.0, 10.0)` was getting stuck...

Got it. I think it's a docs problem. The docs mention the bound for all of the non-central f functions, making it seem like that should be the accepted domain....

For the `mpmath` baselines, I think the best thing to do is use mpmath's [findroot](https://mpmath.org/doc/current/calculus/optimization.html#root-finding-findroot) to invert an mpmath implementation of the cdf function and retry with a higher `dps`...

> OK. Just a double check: this code uses `scipy.special.comb`, written in cython, and hence to the best of my understanding not trivial to use from Pythran. (But the best...

`specfun`actually isn't so bad because each function is mostly self contained, and it can be converted bit by bit. I've already made good headway with `hyp2f1` and that's arguably the...

I'd be happy to review PRs for replacing old Fortran code. I think I can take a look at gh-18570 this week. > I am interested in looking at `dop`...

> thanks for looking into that code! This function is hard to replace as there is no equivalent in boost for example. I have a longer term goal to replace...

> That sounds awesome. Do you think we should still pursue improving the F77 `ive` code here then? To me it looks like a small win that improves the situation...

This could be useful in some cases but I don't think it's that important for `TryExamplesDirective`. I think most users should use autodoc with `global_enable_try_examples = True`, and not bother...