Supplemental support for numerical and mathematical functions
kimikage opened this issue · comments
In response to #179, I checked the support status and performance of some other functions. The (incomplete) list is shown below.
To be supported
-
^(::FixedPoint, ::Integer)
for more accuracy- e.g.
N0f8(0.8)^3 != N0f8(64/125)
- e.g.
-
bitstring
(PR #195) -
big
(Issue #184, PR #186) -
clamp
(Issue #179, PR #194) -
copysign(x::Fixed, y)
andflipsign(x::Fixed, y)
for speed performance (PR #187) -
fma
andmuladd
-
rationalize
(Issue #184, PR #186) - 3-arg
rem
anddiv
(cld
) (PR #226, PR #230) -
rem2pi
andmod2pi
-
signbit
for speed performance (forNormed
) (PR #187)
To be dropped
-
abs(::Normed)
specialization is not helpful. (PR #187)- The fallback implementation works fine.
-
copysign(x::Normed, y)
andflipsign(x::Normed, y)
should be throw error. (PR #187)
TBD
-
maxintfloat
(maxintfixed
andminintfixed
?) -
parse
andtryparse
for fixed point numbers with suffix -
bit shift operators (Issue #250)
- I think they are usuful but somewhat dangerous.
-
numerator
anddenominator
-
widemul
Ah, I forgot widemul
.