pocmatos
2020-4-15 12:31:14
This sounds like a can of worms!
pocmatos
2020-4-15 12:31:37
I don’t have much expertise with floating point arithmetic accuracy. Will take a look at Neil’s paper before commenting any further.
pocmatos
2020-4-15 12:32:28
Also - I feel that once we have a solution that passes, we should get these tests running on racket CI
samth
2020-4-15 14:07:36
@pocmatos I think I figured it out, basically. Neil was just a little over-optimistic about the C library. The issues for the fl2 functions are trickier, but I don’t think anyone but he has expertise to fix them.
pocmatos
2020-4-15 15:08:09
Where’s Neil Toronto these days?
pocmatos
2020-4-15 15:08:32
Was his work in Racket related to the implementation of the math libraries for Racket?
samth
2020-4-15 15:20:57
MSR, and yes.