slow edge cases for `xsf::cephes::detail::hyt2f1`

[lucascolley]

See scipy/scipy#23696 for details. xsf::cephes::detail::hyt2f1 seems to take a long time to produce a NaN for sufficiently large b.

[jorenham]

there's a hypo in hyp2f1

[lucascolley]

there's a hypo in hyp2f1

a hypo or a typo 🤔

[jorenham]

there's a hypo in hyp2f1

a hypo or a typo 🤔

Ah yea sorry, that was a tyto

[lucascolley]

😆 at the risk of communication completely breaking down, on which line is the hypo?

[jorenham]

In the title of this issue :P

[lucascolley]

we might need to switch to full sentences — this issue is about

xsf/include/xsf/cephes/hyp2f1.h

Line 158 in 15ffa02

XSF_HOST_DEVICE double hyt2f1(double a, double b, double c, double x, double *loss) {

[lucascolley]

the chain of x-refs is long, but I narrowed the slowness down to this subroutine (if that is the right word) in scipy/scipy#23654 (comment)

[jorenham]

Oh woops, I had no idea. Sorry bout the noise then :P