negbin fails on X, y data

[awadell1]

Per the doc string: negbin(X::AbstractMatrix, y::AbstractVector, [link::Link]; <keyword arguments>) is valid however:

julia> using GLM

julia> m = negbin(ones(100,1), Int.(rand(UInt32, 100)))
ERROR: type GeneralizedLinearModel has no field model
Stacktrace:
 [1] getproperty(x::GeneralizedLinearModel{GLM.GlmResp{Vector{Float64}, Poisson{Float64}, LogLink}, GLM.DensePredChol{Float64, LinearAlgebra.Cholesky{Float64, Matrix{Float64}}}}, f::Symbol)
   @ Base ./Base.jl:33
 [2] negbin(::Matrix{Float64}, ::Vector{Int64}; initialθ::Float64, maxiter::Int64, minstepfac::Float64, atol::Float64, rtol::Float64, verbose::Bool, kwargs::Base.Iterators.Pairs{Union{}, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})
   @ GLM ~/.julia/packages/GLM/5CcRd/src/negbinfit.jl:112
 [3] negbin(::Matrix{Float64}, ::Vector{Int64})
   @ GLM ~/.julia/packages/GLM/5CcRd/src/negbinfit.jl:78
 [4] top-level scope
   @ REPL[118]:1

Issue seems to stem from negbin assuming that regmodel (Defined at Ln 105/108 in negbinfit.jl ) is a TableRegressionModel and not a GLMResp

[nalimilan]

Good catch. We should indeed check the type of regmodel via something like model = regmodel isa TableRegressionModel ? regmodel.model ? model and then use only model. PR welcome.