StoredArray -> DenseArray for JuliaLang/julia#6258

Eventually we may also work with DArrays and sparse matrices, but as
far as I know, we don't yet.

Author: simonster

src/GLM.jl

@@ -66,7 +66,7 @@ module GLM
         wrkresp         # working response
 
     typealias FP FloatingPoint
-    typealias FPStoredVector{T<:FloatingPoint} StoredArray{T,1}
+    typealias FPVector{T<:FloatingPoint} DenseArray{T,1}
 
     abstract ModResp                   # model response
 

src/glmfit.jl

@@ -1,4 +1,4 @@
-type GlmResp{V<:FPStoredVector} <: ModResp   # response in a glm model
+type GlmResp{V<:FPVector} <: ModResp           # response in a glm model
     y::V                                       # response
     d::UnivariateDistribution
     l::Link
@@ -43,14 +43,14 @@ linkinv!(r::GlmResp) = linkinv!(r.l, r.mu, r.eta)
 # evaluate the mueta vector (derivative of mu w.r.t. eta) from the linear predictor (eta)
 mueta!(r::GlmResp) = mueta!(r.l, r.mueta, r.eta)
 
-function updatemu!{T<:FPStoredVector}(r::GlmResp{T}, linPr::T)
+function updatemu!{T<:FPVector}(r::GlmResp{T}, linPr::T)
     n = length(linPr)
     length(r.offset) == n ? map!(Add(), r.eta, linPr, r.offset) : copy!(r.eta, linPr)
     linkinv!(r); mueta!(r); var!(r); wrkresid!(r); devresid!(r)
     sum(r.devresid)
 end
 
-updatemu!{T<:FPStoredVector}(r::GlmResp{T}, linPr) = updatemu!(r, convert(T,vec(linPr)))
+updatemu!{T<:FPVector}(r::GlmResp{T}, linPr) = updatemu!(r, convert(T,vec(linPr)))
 
 var!(r::GlmResp) = var!(r.d, r.var, r.mu)
 
@@ -142,7 +142,7 @@ function StatsBase.fit(m::GlmMod, y; wts=nothing, offset=nothing, fitargs...)
     dofit ? fit(m; fitargs...) : m
 end
 
-function StatsBase.fit{T<:FloatingPoint,V<:FPStoredVector}(::Type{GlmMod},
+function StatsBase.fit{T<:FloatingPoint,V<:FPVector}(::Type{GlmMod},
                                                            X::Matrix{T}, y::V, d::UnivariateDistribution,
                                                            l::Link=canonicallink(d);
                                                            dofit::Bool=true,

src/lm.jl

@@ -1,4 +1,4 @@
-type LmResp{V<:FPStoredVector} <: ModResp  # response in a linear model
+type LmResp{V<:FPVector} <: ModResp  # response in a linear model
     mu::V                                  # mean response
     offset::V                              # offset added to linear predictor (may have length 0)
     wts::V                                 # prior weights (may have length 0)
@@ -10,14 +10,14 @@ type LmResp{V<:FPStoredVector} <: ModResp  # response in a linear model
         new(mu,off,wts,y)
     end
 end
-LmResp{V<:FPStoredVector}(y::V) = LmResp{V}(fill!(similar(y), zero(eltype(V))), similar(y, 0), similar(y, 0), y)
+LmResp{V<:FPVector}(y::V) = LmResp{V}(fill!(similar(y), zero(eltype(V))), similar(y, 0), similar(y, 0), y)
 
-function updatemu!{V<:FPStoredVector}(r::LmResp{V}, linPr::V)
+function updatemu!{V<:FPVector}(r::LmResp{V}, linPr::V)
     n = length(linPr); length(r.y) == n || error("length(linPr) is $n, should be $(length(r.y))")
     length(r.offset) == 0 ? copy!(r.mu, linPr) : map!(Add(), r.mu, linPr, r.offset)
     deviance(r)
 end
-updatemu!{V<:FPStoredVector}(r::LmResp{V}, linPr) = updatemu!(r, convert(V,vec(linPr)))
+updatemu!{V<:FPVector}(r::LmResp{V}, linPr) = updatemu!(r, convert(V,vec(linPr)))
 
 type WtResid <: Functor{3} end
 evaluate{T<:FP}(::WtResid,wt::T,y::T,mu::T) = (y - mu)*sqrt(wt)