[Bug] Expanding VectorFold with a negative number of folds will (naturally) fail

[jakewilliami]

E.g.,

julia> collect(VectorFold([1, 2], 10, -5))
ERROR: ArgumentError: invalid Array dimensions

Proposed fix: create an internal constructor method for VectorFold (and maybe check PatternFold, as well?) to catch this. Would that be reasonable? Or do we...somehow want non-positive numbers of folds? E.g.:

mutable struct VectorFold{T,V <: AbstractVector{T}, I1 <: Integer, I2 <: Integer} <: PatternFold{T,V}
    pattern::V
    gap::T
    folds::I1
    current::I2
	
    function VectorFold(p::V, g::T, f::I1, c::I2) where {T,V <: AbstractVector{T}, I1 <: Integer, I2 <: Integer}
        f < 0 &&
            throw(ArgumentError("folds cannot be negative (you cannot have a negative number of folds)."))
            new{T, V, I1, I2}(p, g, f, c)
        end
end

[Azzaare]

I can't see a reason to have negative fold either.

You proposition is good. Another way to do it would be to have

mutable struct VectorFold{T,V <: AbstractVector{T}, I1 <: Integer, I2 <: Integer} <: PatternFold{T,V}
    pattern::V
    gap::T
    folds::I1
    current::I2
	
    function VectorFold(p::V, g::T, f::I1, c::I2) where {T,V <: AbstractVector{T}, I1 <: Integer, I2 <: Integer}
            new{T, V, I1, I2}(p, g, max(0,f), c)
        end
end

But it is prone to introduce unexpected behavior. I would rather have your solution (maybe with the following variant)

mutable struct VectorFold{T,V <: AbstractVector{T}, I1 <: Integer, I2 <: Integer} <: PatternFold{T,V}
    pattern::V
    gap::T
    folds::I1
    current::I2
	
    function VectorFold(p::V, g::T, f::I1, c::I2) where {T,V <: AbstractVector{T}, I1 <: Integer, I2 <: Integer}
            @assert f => 0
            new{T, V, I1, I2}(p, g, f, c)
    end
end