Bump minimum Julia version to 0.6 (#22)

.travis.yml

@@ -1,16 +1,17 @@
 language: julia
 os:
   - linux
+addons:
+  apt:
+    packages:
+      - gfortran
 julia:
-  - 0.4
-  - 0.5
+  - 0.6
   - nightly
 notifications:
   email: false
-before_install:
-  - sudo apt-get install gfortran -y
-script:
-  - if [[ -a .git/shallow ]]; then git fetch --unshallow; fi
-  - julia --check-bounds=yes -e 'Pkg.clone(pwd()); Pkg.build("GLMNet"); Pkg.test("GLMNet"; coverage=true)'
+#script:
+#  - if [[ -a .git/shallow ]]; then git fetch --unshallow; fi
+#  - julia --check-bounds=yes -e 'Pkg.clone(pwd()); Pkg.build("GLMNet"); Pkg.test("GLMNet"; coverage=true)'
 after_success:
-  - if [ $TRAVIS_JULIA_VERSION = "nightly" ]; then julia -e 'cd(Pkg.dir("GLMNet")); Pkg.add("Coverage"); using Coverage; Coveralls.submit(Coveralls.process_folder())'; fi
+  - julia -e 'cd(Pkg.dir("GLMNet")); Pkg.add("Coverage"); using Coverage; Coveralls.submit(Coveralls.process_folder())'

README.md

@@ -1,7 +1,7 @@
 # GLMNet
 
-[![Build Status](https://travis-ci.org/simonster/GLMNet.jl.svg?branch=master)](https://travis-ci.org/simonster/GLMNet.jl)
-[![Coverage Status](https://coveralls.io/repos/simonster/GLMNet.jl/badge.svg?branch=master)](https://coveralls.io/r/simonster/GLMNet.jl?branch=master)
+[![Build Status](https://travis-ci.org/JuliaStats/GLMNet.jl.svg?branch=v0.0.4)](https://travis-ci.org/JuliaStats/GLMNet.jl)
+[![Coverage Status](https://coveralls.io/repos/github/JuliaStats/GLMNet.jl/badge.svg)](https://coveralls.io/github/JuliaStats/GLMNet.jl)
 
 [glmnet](http://www.jstatsoft.org/v33/i01/) is an R package by Jerome Friedman, Trevor Hastie, Rob Tibshirani that fits entire Lasso or ElasticNet regularization paths for linear, logistic, multinomial, and Cox models using cyclic coordinate descent. This Julia package wraps the Fortran code from glmnet.
 
@@ -34,9 +34,9 @@ Least Squares GLMNet Solution Path (55 solutions for 4 predictors in 163 passes)
 ```julia
 julia> path.betas
 4x55 CompressedPredictorMatrix:
- 0.0  0.083706  0.159976  0.22947  …  0.929157    0.929315  
+ 0.0  0.083706  0.159976  0.22947  …  0.929157    0.929315
  0.0  0.0       0.0       0.0         0.00655753  0.00700862
- 0.0  0.0       0.0       0.0         0.0         0.0       
+ 0.0  0.0       0.0       0.0         0.0         0.0
  0.0  0.0       0.0       0.0         0.0         0.0
 ```
 
@@ -63,9 +63,9 @@ julia> indmin(cv.meanloss)
 
 julia> cv.path.betas[:, 48]
 4-element Array{Float64,1}:
- 0.926911  
+ 0.926911
  0.00366805
- 0.0       
+ 0.0
  0.0
 ```
 

REQUIRE

@@ -1,4 +1,4 @@
-julia 0.4
+julia 0.6
 Distributions
 StatsBase
 Compat 0.9.3

deps/build.jl

@@ -1,5 +1,5 @@
 using Compat
-@static if is_windows()
+if Compat.Sys.iswindows()
     flags = ["-m$(Sys.WORD_SIZE)","-fdefault-real-8","-ffixed-form","-shared","-O3"]
 else
     flags = ["-m$(Sys.WORD_SIZE)","-fdefault-real-8","-ffixed-form","-shared","-O3","-fPIC"]

src/GLMNet.jl

@@ -1,3 +1,5 @@
+__precompile__()
+
 module GLMNet
 using Distributions, Compat, StatsBase
 
@@ -47,21 +49,15 @@ function getindex(X::CompressedPredictorMatrix, a::AbstractVector{Int}, b::Abstr
 end
 
 
-if VERSION < v"0.5.0"
-    # use old slicing behaviour
-    getindex(X::CompressedPredictorMatrix, a::Int, b::AbstractVector{Int}) = getindex(X,[a],b)
-else
-    # use new slicing behaviour
-    function getindex(X::CompressedPredictorMatrix, a::Int, b::AbstractVector{Int})
-        checkbounds(X, a, b)
-        out = zeros(length(b))
-        for j = 1:length(b), i = 1:X.nin[b[j]]
-            if a == X.ia[i]
-                out[j] = X.ca[i, b[j]]
-            end
+function getindex(X::CompressedPredictorMatrix, a::Int, b::AbstractVector{Int})
+    checkbounds(X, a, b)
+    out = zeros(length(b))
+    for j = 1:length(b), i = 1:X.nin[b[j]]
+        if a == X.ia[i]
+            out[j] = X.ca[i, b[j]]
         end
-        out
     end
+    out
 end
 
 # Get number of active predictors for a model in X
@@ -76,7 +72,7 @@ end
 nactive(X::CompressedPredictorMatrix, b::AbstractVector{Int}=1:length(X.nin)) =
     [nactive(X, j) for j in b]
 
-function convert{T<:Matrix{Float64}}(::Type{T}, X::CompressedPredictorMatrix)
+function convert(::Type{Matrix{Float64}}, X::CompressedPredictorMatrix)
     mat = zeros(X.ni, length(X.nin))
     for b = 1:size(mat, 2), i = 1:X.nin[b]
         mat[X.ia[i], b] = X.ca[i, b]