gamutmax(LCHab) disagrees with the Colors.jl usage

[kimikage]

The max saturation of LCHab is defined as 1 in ColorTypes.jl.

ColorTypes.jl/src/operations.jl

Lines 16 to 17 in 0d5904e

	gamutmax(::Type{T}) where {T<:LCHab} = (100,1,360)
	gamutmin(::Type{T}) where {T<:LCHab} = (0,0,0)

However, in Colors.jl, the saturation comes from the a and b of Lab which are mostly in [-100,100].
https://github.com/JuliaGraphics/Colors.jl/blob/657e265a4cccc9fc32584f7f3505611537bfc79d/src/conversions.jl#L575-L580

Therefore, for the sake of convenience, 100 (or 128) is better for the max saturation, though the actual gamut is not so simple. (See also: #84)

[timholy]

Another option is to remove support for ones without a clearly defined max/min?

[kimikage]

In the case of LCHab, the removal is probably OK, as no one has pointed this out.

However, few types have a clearly defined max/min, because the range of color space and the color gamut are based on different concepts.

BTW, I found this issue when I ran rand(LCHab).
I think gamutmax and gamutmin are practical when and only when they are used with rand() as historically they are introduced for such a purpose.
Perhaps what we should remove might be the term “gamut” rather than the function.

[kimikage]

As mentioned above, I think that "color gamut" and rand should be separated.
Then, rand should be more comprehensive. For example, there is a problem with TransparentColor:

julia> rand(RGBA)
RGBA{Float64}(0.4054196785489861,0.5812337734513682,0.4739237429262706,0.35592317884458446)

julia> rand(Lab)
Lab{Float64}(49.35707185830977,54.78932461097497,-80.6404236285697)

julia> rand(LabA)
ERROR: MethodError: no method matching gamutmax(::Type{LabA{Float64}})

Although it is necessary to discuss whether ColorTypes.jl should specify the ranges (not gamuts) or not, it may be the duty of ColorTypes.jl to resolve the range of AlphaColor and ColorAlpha, given the range of color_type.

[kimikage]

BTW, gamutmin/gamutmax for YIQ are broken, too.

ColorTypes.jl/src/operations.jl

Lines 18 to 19 in 9607b73

	gamutmax(::Type{T}) where {T<:YIQ} = (1,0.5226,0.5226)
	gamutmin(::Type{T}) where {T<:YIQ} = (0,-0.5957,-0.5957)

correct_gamut(c::YIQ{T}) where {T} = YIQ{T}(clamp(c.y, zero(T), one(T)),
                                     clamp(c.i, convert(T,-0.5957), convert(T,0.5957)),
                                     clamp(c.q, convert(T,-0.5226), convert(T,0.5226)))