2 move ldouble manipulation into arch specific headers
3 long double isnan, isinf..:
4 efficient macro versions of long double classification macros?
5 (they are ld* representation specific)
7 does inexact allowed on overflow?
8 see c99 7.12, 7.12.9.5, F.9, F.9.6.5
10 if libm.h internals is not needed (ldexp,..) then include math.h only
11 LD64 wrap small long double functions as well
12 trigonometric functions:
13 __rem_pio_large: is returning 2 bits enough?
14 __tan: 3rd arg semantics is probably not optimal
15 sinf,cosf: return sindf(-y) vs -sindf(y)?
16 __sin,__cos,..: z,w,.. -> x2,x4,.. so degree is easier to see
17 __rem_pio2*: rounding trick: simply use the low 32 bits of fn?
19 use strtod? c99 7.12.11.2
21 scalb is buggy, do we need the *f and *l version?
23 needs a rewrite: signed int problems, wrong <-50000 check
25 only complex in math/, move to cmath?
27 declare long double bessel functions in math.h?
28 uniform inexact exception raising:
29 (int)x == 0 and huge+x > 0 are used
30 unreachable code is not commented and hard to follow
32 int32_t -> uint32_t conversion (subtle)
34 TWO52, one, zero, twom1000,.. -> use reasonable consts
36 ldexp -> use scalbn internally (ldexp is a wrapper)
37 i386 vs precision -> assume extended prec setting
38 long double const trunc/slowness bug
39 remove "unsupported long double format" checks
42 expects single 53bit rounding
43 rewrite using long double arithmetics?
45 dekker vs long double arithmetics
48 lanczos approx as in boost/math/special_functions, python/Modules/mathmodule.c
51 optimizable creal cimag (libm.h macro for internal code?)
52 test? (using mdc lib, available test vectors,..)
53 Kahan, W. "Branch Cuts for Complex Elementary Functions, or Much Ado About Nothing's Sign Bit." 1987
54 Hull, Fairgrieve, Tang "Implementing complex elementary functions using exception handling" 1994
55 Hull, Fairgrieve, Tang "Implementing the complex arcsine and arccosine functions using exception handling" 1997
57 python/Modules/cmathmodule.c (+math, cmath tests)
59 i386, x86_64 asm versions
62 special cases from c99 F.9, G
63 randomized test using mpfr