Stop using deprecated signatures from Distances.jl (#529)

Author: simsurace

Project.toml

@@ -1,6 +1,6 @@
 name = "KernelFunctions"
 uuid = "ec8451be-7e33-11e9-00cf-bbf324bd1392"
-version = "0.10.57"
+version = "0.10.58"
 
 [deps]
 ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"
@@ -25,7 +25,7 @@ ZygoteRules = "700de1a5-db45-46bc-99cf-38207098b444"
 ChainRulesCore = "1"
 Compat = "3.7, 4"
 CompositionsBase = "0.1"
-Distances = "0.10"
+Distances = "0.10.9"
 FillArrays = "0.10, 0.11, 0.12, 0.13, 1"
 Functors = "0.1, 0.2, 0.3, 0.4"
 IrrationalConstants = "0.1, 0.2"

src/basekernels/fbm.jl

@@ -57,7 +57,7 @@ end
 
 function kernelmatrix!(K::AbstractMatrix, κ::FBMKernel, x::AbstractVector)
     modx = _mod(x)
-    pairwise!(K, SqEuclidean(), x)
+    pairwise!(SqEuclidean(), K, x)
     K .= _fbm.(modx, modx', K, κ.h)
     return K
 end
@@ -72,7 +72,7 @@ end
 function kernelmatrix!(
     K::AbstractMatrix, κ::FBMKernel, x::AbstractVector, y::AbstractVector
 )
-    pairwise!(K, SqEuclidean(), x, y)
+    pairwise!(SqEuclidean(), K, x, y)
     K .= _fbm.(_mod(x), _mod(y)', K, κ.h)
     return K
 end

src/distances/pairwise.jl

@@ -6,11 +6,11 @@ end
 
 pairwise(d::PreMetric, X::AbstractVector) = pairwise(d, X, X)
 
-function pairwise!(out::AbstractMatrix, d::PreMetric, X::AbstractVector, Y::AbstractVector)
+function pairwise!(d::PreMetric, out::AbstractMatrix, X::AbstractVector, Y::AbstractVector)
     return broadcast!(d, out, X, permutedims(Y))
 end
 
-pairwise!(out::AbstractMatrix, d::PreMetric, X::AbstractVector) = pairwise!(out, d, X, X)
+pairwise!(d::PreMetric, out::AbstractMatrix, X::AbstractVector) = pairwise!(d, out, X, X)
 
 function pairwise(d::PreMetric, x::AbstractVector{<:Real})
     return Distances_pairwise(d, reshape(x, :, 1); dims=1)
@@ -20,14 +20,14 @@ function pairwise(d::PreMetric, x::AbstractVector{<:Real}, y::AbstractVector{<:R
     return Distances_pairwise(d, reshape(x, :, 1), reshape(y, :, 1); dims=1)
 end
 
-function pairwise!(out::AbstractMatrix, d::PreMetric, x::AbstractVector{<:Real})
-    return Distances.pairwise!(out, d, reshape(x, :, 1); dims=1)
+function pairwise!(d::PreMetric, out::AbstractMatrix, x::AbstractVector{<:Real})
+    return Distances.pairwise!(d, out, reshape(x, :, 1); dims=1)
 end
 
 function pairwise!(
-    out::AbstractMatrix, d::PreMetric, x::AbstractVector{<:Real}, y::AbstractVector{<:Real}
+    d::PreMetric, out::AbstractMatrix, x::AbstractVector{<:Real}, y::AbstractVector{<:Real}
 )
-    return Distances.pairwise!(out, d, reshape(x, :, 1), reshape(y, :, 1); dims=1)
+    return Distances.pairwise!(d, out, reshape(x, :, 1), reshape(y, :, 1); dims=1)
 end
 
 # Also defines the colwise method for abstractvectors

src/matrix/kernelmatrix.jl

@@ -134,15 +134,15 @@ kernelmatrix_diag(κ::Kernel, x::AbstractVector, y::AbstractVector) = map(κ, x,
 
 function kernelmatrix!(K::AbstractMatrix, κ::SimpleKernel, x::AbstractVector)
     validate_inplace_dims(K, x)
-    pairwise!(K, metric(κ), x)
+    pairwise!(metric(κ), K, x)
     return map!(x -> kappa(κ, x), K, K)
 end
 
 function kernelmatrix!(
     K::AbstractMatrix, κ::SimpleKernel, x::AbstractVector, y::AbstractVector
 )
     validate_inplace_dims(K, x, y)
-    pairwise!(K, metric(κ), x, y)
+    pairwise!(metric(κ), K, x, y)
     return map!(x -> kappa(κ, x), K, K)
 end
 

src/utils.jl

@@ -107,11 +107,11 @@ end
 function pairwise(d::PreMetric, x::ColVecs, y::AbstractVector{<:AbstractVector{<:Real}})
     return Distances_pairwise(d, x.X, reduce(hcat, y); dims=2)
 end
-function pairwise!(out::AbstractMatrix, d::PreMetric, x::ColVecs)
-    return Distances.pairwise!(out, d, x.X; dims=2)
+function pairwise!(d::PreMetric, out::AbstractMatrix, x::ColVecs)
+    return Distances.pairwise!(d, out, x.X; dims=2)
 end
-function pairwise!(out::AbstractMatrix, d::PreMetric, x::ColVecs, y::ColVecs)
-    return Distances.pairwise!(out, d, x.X, y.X; dims=2)
+function pairwise!(d::PreMetric, out::AbstractMatrix, x::ColVecs, y::ColVecs)
+    return Distances.pairwise!(d, out, x.X, y.X; dims=2)
 end
 
 """
@@ -178,11 +178,11 @@ end
 function pairwise(d::PreMetric, x::RowVecs, y::AbstractVector{<:AbstractVector{<:Real}})
     return Distances_pairwise(d, x.X, permutedims(reduce(hcat, y)); dims=1)
 end
-function pairwise!(out::AbstractMatrix, d::PreMetric, x::RowVecs)
-    return Distances.pairwise!(out, d, x.X; dims=1)
+function pairwise!(d::PreMetric, out::AbstractMatrix, x::RowVecs)
+    return Distances.pairwise!(d, out, x.X; dims=1)
 end
-function pairwise!(out::AbstractMatrix, d::PreMetric, x::RowVecs, y::RowVecs)
-    return Distances.pairwise!(out, d, x.X, y.X; dims=1)
+function pairwise!(d::PreMetric, out::AbstractMatrix, x::RowVecs, y::RowVecs)
+    return Distances.pairwise!(d, out, x.X, y.X; dims=1)
 end
 
 # Resolve ambiguity error for ColVecs vs RowVecs. #346

test/distances/pairwise.jl

@@ -11,10 +11,10 @@
 
     @test KernelFunctions.pairwise(d, x, y) ≈ pairwise(d, X, Y; dims=2)
     @test KernelFunctions.pairwise(d, x) ≈ pairwise(d, X; dims=2)
-    KernelFunctions.pairwise!(K, d, x, y)
+    KernelFunctions.pairwise!(d, K, x, y)
     @test K ≈ pairwise(d, X, Y; dims=2)
     K = zeros(Ns[1], Ns[1])
-    KernelFunctions.pairwise!(K, d, x)
+    KernelFunctions.pairwise!(d, K, x)
     @test K ≈ pairwise(d, X; dims=2)
 
     x = randn(rng, 10)
@@ -24,6 +24,6 @@
     K = zeros(10, 11)
     @test KernelFunctions.pairwise(d, x, y) ≈ pairwise(d, X, Y; dims=1)
     @test KernelFunctions.pairwise(d, x) ≈ pairwise(d, X; dims=1)
-    KernelFunctions.pairwise!(K, d, x, y)
+    KernelFunctions.pairwise!(d, K, x, y)
     @test K ≈ pairwise(d, X, Y; dims=1)
 end

test/utils.jl

@@ -47,10 +47,10 @@
         @test vcat(DX, DY) isa ColVecs
         @test vcat(DX, DY).X == hcat(X, Y)
         K = zeros(N, N)
-        KernelFunctions.pairwise!(K, SqEuclidean(), DX)
+        KernelFunctions.pairwise!(SqEuclidean(), K, DX)
         @test K ≈ pairwise(SqEuclidean(), X; dims=2)
         K = zeros(N, N + 1)
-        KernelFunctions.pairwise!(K, SqEuclidean(), DX, DY)
+        KernelFunctions.pairwise!(SqEuclidean(), K, DX, DY)
         @test K ≈ pairwise(SqEuclidean(), X, Y; dims=2)
 
         let
@@ -105,10 +105,10 @@
         @test vcat(DX, DY) isa RowVecs
         @test vcat(DX, DY).X == vcat(X, Y)
         K = zeros(D, D)
-        KernelFunctions.pairwise!(K, SqEuclidean(), DX)
+        KernelFunctions.pairwise!(SqEuclidean(), K, DX)
         @test K ≈ pairwise(SqEuclidean(), X; dims=1)
         K = zeros(D, D + 1)
-        KernelFunctions.pairwise!(K, SqEuclidean(), DX, DY)
+        KernelFunctions.pairwise!(SqEuclidean(), K, DX, DY)
         @test K ≈ pairwise(SqEuclidean(), X, Y; dims=1)
 
         let