2 drop ld128 support? and move ldshape union to arch/
3 (at least rename ieeel2 union and use the same union in all *l.c)
5 -ffloat-store or -fexcess-precision=standard probably fixes most volatile issues (and strict_assign)
6 volatile const may need different treatment
8 floor, ceil.. probably can be improved
10 nan 0/0 raises invalid
13 uniform inexact/invalid exception raising:
14 (int)x == 0 and huge+x > 0 are used
15 (unreachable code is not commented and hard to follow)
17 dekker vs long double arithmetics
19 move ldouble manipulation into arch specific headers
20 long double isnan, isinf..:
21 efficient macro versions of long double classification macros?
22 (they are ld* representation specific)
24 no inexact exception (or fix lround)
27 trigonometric functions:
28 __rem_pio_large: is returning 2 bits enough?
29 __tan: 3rd arg semantics is probably not optimal
30 sinf,cosf: return sindf(-y) vs -sindf(y)?
31 __sin,__cos,..: z,w,.. -> x2,x4,.. so degree is easier to see
32 __rem_pio2*: rounding trick: simply use the low 32 bits of fn?
33 use long double pi in long double code? (casin, cacos,..)
35 scalb is buggy, do we need the *f and *l version?
37 int32_t -> uint32_t conversion (subtle)
39 TWO52, twom1000 vs tiny
40 remove "unsupported long double format" checks
41 remove overflow thresholds (sinh, cosh) when result overflows anyway?
42 sign bit checking convention (sqrt.c)
49 lanczos approx as in boost/math/special_functions and python/Modules/mathmodule.c
52 optimizable creal cimag (libm.h macro for internal code?)
53 include <math.h> and <complex.h> instead of libm.h (once there are efficient creal etc.)
54 cpack(x,y) vs x+I*y vs union .a[0]=x, .a[1]=y
55 fix casin[h], cacos[h], catan[h] (complex arith cornercases)
56 add missing long double versions
57 Kahan, W. "Branch Cuts for Complex Elementary Functions, or Much Ado About Nothing's Sign Bit." 1987
58 Hull, Fairgrieve, Tang "Implementing complex elementary functions using exception handling" 1994
59 Hull, Fairgrieve, Tang "Implementing the complex arcsine and arccosine functions using exception handling" 1997
61 python/Modules/cmathmodule.c
65 test vectors (see if python has good math+cmath tests)
66 special cases from c99 F.9, G.6
67 randomized test using mpfr and mdc for complex
73 code depends on signed int right shift behaviour
75 pow*(1,nan or inf) == 1 special case
78 fma* x == 0 should be after the isfinite check
79 invalid ldexp optimizaion
80 lrint* inexact clearing was optimized away
81 scalbn(0x1p1023, -2097)==0x1p-1074 was incorrect
82 ilogbl was incorrect on ld80 (implicit bit was not taken into account)