can not convert float to the same or more number of bit fixed point

[xiuliren]

this works in Julia 0.6.2:

Float32(1.0) |> N0f8
Float32(1.0) |> N0f16

this will not work

Float32(1.0) |> N0f32
Float32(1.0) |> N0f64
Float64(1.0) |> N0f64

It looks like when the number of bits is the same or larger, it won’t work.
error message:

ArgumentError: FixedPointNumbers.Normed{UInt64,64} is a 64-bit type representing 0 values from 0.0 to 1.0; cannot represent 1.0

[giordano]

I just run into the same error. The problem is in _convert(::Type{U}, ::Type{T}, x) where {U <: Normed,T} function:

FixedPointNumbers.jl/src/normed.jl

Lines 52 to 55 in cf9ebc4

	function _convert(::Type{U}, ::Type{T}, x) where {U <: Normed,T}
	y = round(widen1(rawone(U))*x)
	(0 <= y) & (y <= typemax(T)) || throw_converterror(U, x)
	U(_unsafe_trunc(T, y), 0)

E.g., in

FixedPointNumbers._convert(N0f64, FixedPointNumbers.rawtype(N0f64), 1.0)

the y variable is equal to

julia> y = round(FixedPointNumbers.widen1(FixedPointNumbers.rawone(N0f64)) * 1.0)
1.8446744073709552e19

which is larger than the maximum number that can be represented with UInt64:

julia> big(typemax(UInt64))
18446744073709551615

Maybe it's worth checking if the x argument is one?

[kimikage]

Maybe it's worth checking if the x argument is one?

FYI, since the length of the mantissa part of Float32 is 23 (+1) bits, this is not a problem just when x argument is one.

julia> for f=1:32; Float32(exp2(32-f)) |> Normed{UInt32,f};end
ERROR: ArgumentError: Normed{UInt32,25} is a 32-bit type representing 4294967296 values from 0.0 to 128.0; cannot represent 128.0

Similarly, the length of the mantissa part of Float64 is 52 (+1) bits.

julia> for f=1:64; Float64(exp2(64-f)) |> Normed{UInt64,f};end
ERROR: ArgumentError: Normed{UInt64,54} is a 64-bit type representing 0 values from 0.0 to 1024.0; cannot represent 1024.0

Fixing this problem may be useful for the round trip testing to check the conversions.

[kimikage]

PR #131 does not change the rem.

Therefore, this does not fix the problems such as:

julia> using Colors

julia> white = parse(RGB{Float32}, "white")
RGB{Float32}(1.0f0,1.0f0,1.0f0)

julia> convert(RGB{N0f32}, white)
RGB{N0f32}(0.0,0.0,0.0)

Originally posted by @kimikage in #131 (comment)

When you are faced by a problem with rem, please report it as a new issue.