fix: replace copyto! with copy!

.github/workflows/Test.yml

@@ -25,7 +25,7 @@ jobs:
       actions: write
       contents: read
     strategy:
-      fail-fast: true # TODO: toggle
+      fail-fast: false # TODO: toggle
       matrix:
         version:
           - '1.10'

DifferentiationInterface/ext/DifferentiationInterfaceEnzymeExt/forward_onearg.jl

@@ -107,7 +107,7 @@ function DI.value_and_pushforward!(
     DI.check_prep(f, prep, backend, x, tx, contexts...)
     # dy cannot be passed anyway
     y, new_ty = DI.value_and_pushforward(f, prep, backend, x, tx, contexts...)
-    foreach(copyto!, ty, new_ty)
+    foreach(copy!, ty, new_ty)
     return y, ty
 end
 
@@ -123,7 +123,7 @@ function DI.pushforward!(
     DI.check_prep(f, prep, backend, x, tx, contexts...)
     # dy cannot be passed anyway
     new_ty = DI.pushforward(f, prep, backend, x, tx, contexts...)
-    foreach(copyto!, ty, new_ty)
+    foreach(copy!, ty, new_ty)
     return ty
 end
 
@@ -198,7 +198,7 @@ function DI.gradient!(
         contexts::Vararg{DI.Constant, C},
     ) where {F, SIG, B, C}
     DI.check_prep(f, prep, backend, x, contexts...)
-    return copyto!(grad, DI.gradient(f, prep, backend, x, contexts...))
+    return copy!(grad, DI.gradient(f, prep, backend, x, contexts...))
 end
 
 function DI.value_and_gradient!(
@@ -211,7 +211,7 @@ function DI.value_and_gradient!(
     ) where {F, SIG, B, C}
     DI.check_prep(f, prep, backend, x, contexts...)
     y, new_grad = DI.value_and_gradient(f, prep, backend, x, contexts...)
-    return y, copyto!(grad, new_grad)
+    return y, copy!(grad, new_grad)
 end
 
 ## Jacobian
@@ -291,7 +291,7 @@ function DI.jacobian!(
         contexts::Vararg{DI.Constant, C},
     ) where {F, C}
     DI.check_prep(f, prep, backend, x, contexts...)
-    return copyto!(jac, DI.jacobian(f, prep, backend, x, contexts...))
+    return copy!(jac, DI.jacobian(f, prep, backend, x, contexts...))
 end
 
 function DI.value_and_jacobian!(
@@ -304,5 +304,5 @@ function DI.value_and_jacobian!(
     ) where {F, C}
     DI.check_prep(f, prep, backend, x, contexts...)
     y, new_jac = DI.value_and_jacobian(f, prep, backend, x, contexts...)
-    return y, copyto!(jac, new_jac)
+    return y, copy!(jac, new_jac)
 end

DifferentiationInterface/ext/DifferentiationInterfaceEnzymeExt/reverse_twoarg.jl

@@ -35,7 +35,7 @@ function DI.value_and_pullback(
     ) where {F, C}
     DI.check_prep(f!, y, prep, backend, x, ty, contexts...)
     (; df!, context_shadows, ty_copy) = prep
-    copyto!(only(ty_copy), only(ty))
+    copy!(only(ty_copy), only(ty))
     mode = reverse_noprimal(backend)
     f!_and_df! = get_f_and_df_prepared!(df!, f!, backend, Val(1))
     dy = only(ty_copy)
@@ -59,7 +59,7 @@ function DI.value_and_pullback(
     ) where {F, B, C}
     DI.check_prep(f!, y, prep, backend, x, ty, contexts...)
     (; df!, context_shadows, ty_copy) = prep
-    foreach(copyto!, ty_copy, ty)
+    foreach(copy!, ty_copy, ty)
     mode = reverse_noprimal(backend)
     f!_and_df! = get_f_and_df_prepared!(df!, f!, backend, Val(B))
     ty = ty_copy
@@ -83,7 +83,7 @@ function DI.value_and_pullback(
     ) where {F, C}
     DI.check_prep(f!, y, prep, backend, x, ty, contexts...)
     (; df!, context_shadows, ty_copy) = prep
-    copyto!(only(ty_copy), only(ty))
+    copy!(only(ty_copy), only(ty))
     mode = reverse_noprimal(backend)
     f!_and_df! = get_f_and_df_prepared!(df!, f!, backend, Val(1))
     dx = make_zero(x)  # allocates
@@ -106,7 +106,7 @@ function DI.value_and_pullback(
     ) where {F, B, C}
     DI.check_prep(f!, y, prep, backend, x, ty, contexts...)
     (; df!, context_shadows, ty_copy) = prep
-    foreach(copyto!, ty_copy, ty)
+    foreach(copy!, ty_copy, ty)
     mode = reverse_noprimal(backend)
     f!_and_df! = get_f_and_df_prepared!(df!, f!, backend, Val(B))
     tx = ntuple(_ -> make_zero(x), Val(B))  # allocates
@@ -130,7 +130,7 @@ function DI.value_and_pullback!(
     ) where {F, C}
     DI.check_prep(f!, y, prep, backend, x, ty, contexts...)
     (; df!, context_shadows, ty_copy) = prep
-    copyto!(only(ty_copy), only(ty))
+    copy!(only(ty_copy), only(ty))
     mode = reverse_noprimal(backend)
     f!_and_df! = get_f_and_df_prepared!(df!, f!, backend, Val(1))
     dx = only(tx)
@@ -155,7 +155,7 @@ function DI.value_and_pullback!(
     ) where {F, B, C}
     DI.check_prep(f!, y, prep, backend, x, ty, contexts...)
     (; df!, context_shadows, ty_copy) = prep
-    foreach(copyto!, ty_copy, ty)
+    foreach(copy!, ty_copy, ty)
     mode = reverse_noprimal(backend)
     f!_and_df! = get_f_and_df_prepared!(df!, f!, backend, Val(B))
     make_zero!(tx)

DifferentiationInterface/ext/DifferentiationInterfaceFiniteDiffExt/onearg.jl

@@ -403,7 +403,7 @@ function DI.jacobian!(
     DI.check_prep(f, prep, backend, x, contexts...)
     (; relstep, absstep, dir) = prep
     fc = DI.fix_tail(f, map(DI.unwrap, contexts)...)
-    return copyto!(
+    return copy!(
         jac,
         finite_difference_jacobian(
             fc, x, prep.cache; jac_prototype = jac, relstep, absstep, dir
@@ -425,7 +425,7 @@ function DI.value_and_jacobian!(
     y = fc(x)
     return (
         y,
-        copyto!(
+        copy!(
             jac,
             finite_difference_jacobian(
                 fc, x, prep.cache, y; jac_prototype = jac, relstep, absstep, dir

DifferentiationInterface/ext/DifferentiationInterfaceFiniteDifferencesExt/DifferentiationInterfaceFiniteDifferencesExt.jl

@@ -136,7 +136,7 @@ function DI.gradient!(
         contexts::Vararg{DI.Context, C},
     ) where {C}
     DI.check_prep(f, prep, backend, x, contexts...)
-    return copyto!(grad, DI.gradient(f, prep, backend, x, contexts...))
+    return copy!(grad, DI.gradient(f, prep, backend, x, contexts...))
 end
 
 function DI.value_and_gradient!(
@@ -149,7 +149,7 @@ function DI.value_and_gradient!(
     ) where {C}
     DI.check_prep(f, prep, backend, x, contexts...)
     y, new_grad = DI.value_and_gradient(f, prep, backend, x, contexts...)
-    return y, copyto!(grad, new_grad)
+    return y, copy!(grad, new_grad)
 end
 
 ## Jacobian
@@ -193,7 +193,7 @@ function DI.jacobian!(
         contexts::Vararg{DI.Context, C},
     ) where {C}
     DI.check_prep(f, prep, backend, x, contexts...)
-    return copyto!(jac, DI.jacobian(f, prep, backend, x, contexts...))
+    return copy!(jac, DI.jacobian(f, prep, backend, x, contexts...))
 end
 
 function DI.value_and_jacobian!(
@@ -206,7 +206,7 @@ function DI.value_and_jacobian!(
     ) where {C}
     DI.check_prep(f, prep, backend, x, contexts...)
     y, new_jac = DI.value_and_jacobian(f, prep, backend, x, contexts...)
-    return y, copyto!(jac, new_jac)
+    return y, copy!(jac, new_jac)
 end
 
 end

DifferentiationInterface/ext/DifferentiationInterfaceForwardDiffExt/onearg.jl

@@ -298,7 +298,7 @@ function DI.value_and_gradient!(
         result = DiffResult(zero(eltype(x)), (grad,))
         result = gradient!(result, fc, x)
         y = DR.value(result)
-        grad === DR.gradient(result) || copyto!(grad, DR.gradient(result))
+        grad === DR.gradient(result) || copy!(grad, DR.gradient(result))
         return y, grad
     else
         prep = DI.prepare_gradient_nokwarg(Val(true), f, backend, x, contexts...)
@@ -397,7 +397,7 @@ function DI.value_and_gradient!(
     end
     result = gradient!(result, fc, x, prep.config, Val(false))
     y = DR.value(result)
-    grad === DR.gradient(result) || copyto!(grad, DR.gradient(result))
+    grad === DR.gradient(result) || copy!(grad, DR.gradient(result))
     return y, grad
 end
 
@@ -472,7 +472,7 @@ function DI.value_and_jacobian!(
         result = DiffResult(y, (jac,))
         result = jacobian!(result, fc, x)
         y = DR.value(result)
-        jac === DR.jacobian(result) || copyto!(jac, DR.jacobian(result))
+        jac === DR.jacobian(result) || copy!(jac, DR.jacobian(result))
         return y, jac
     else
         prep = DI.prepare_jacobian_nokwarg(Val(true), f, backend, x, contexts...)
@@ -566,7 +566,7 @@ function DI.value_and_jacobian!(
     end
     result = jacobian!(result, fc, x, prep.config, Val(false))
     y = DR.value(result)
-    jac === DR.jacobian(result) || copyto!(jac, DR.jacobian(result))
+    jac === DR.jacobian(result) || copy!(jac, DR.jacobian(result))
     return y, jac
 end
 
@@ -761,8 +761,8 @@ function DI.value_gradient_and_hessian!(
         result = DiffResult(oneunit(eltype(x)), (grad, hess))
         result = hessian!(result, fc, x)
         y = DR.value(result)
-        grad === DR.gradient(result) || copyto!(grad, DR.gradient(result))
-        hess === DR.hessian(result) || copyto!(hess, DR.hessian(result))
+        grad === DR.gradient(result) || copy!(grad, DR.gradient(result))
+        hess === DR.hessian(result) || copy!(hess, DR.hessian(result))
         return (y, grad, hess)
     else
         prep = DI.prepare_hessian_nokwarg(Val(true), f, backend, x, contexts...)
@@ -864,8 +864,8 @@ function DI.value_gradient_and_hessian!(
     end
     result = hessian!(result, fc, x, prep.result_config, Val(false))
     y = DR.value(result)
-    grad === DR.gradient(result) || copyto!(grad, DR.gradient(result))
-    hess === DR.hessian(result) || copyto!(hess, DR.hessian(result))
+    grad === DR.gradient(result) || copy!(grad, DR.gradient(result))
+    hess === DR.hessian(result) || copy!(hess, DR.hessian(result))
     return (y, grad, hess)
 end
 

DifferentiationInterface/ext/DifferentiationInterfaceMooncakeExt/forward_onearg.jl

@@ -78,7 +78,7 @@ function DI.value_and_pushforward!(
     ) where {F, C}
     DI.check_prep(f, prep, backend, x, tx, contexts...)
     y, new_ty = DI.value_and_pushforward(f, prep, backend, x, tx, contexts...)
-    foreach(copyto!, ty, new_ty)
+    foreach(copy!, ty, new_ty)
     return y, ty
 end
 

DifferentiationInterface/ext/DifferentiationInterfaceMooncakeExt/forward_twoarg.jl

@@ -99,7 +99,7 @@ function DI.value_and_pushforward!(
             Dual(x, dx_righttype),
             map(Dual_unwrap, contexts, prep.context_tangents)...,
         )
-        dy === dy_righttype || copyto!(dy, dy_righttype)
+        dy === dy_righttype || copy!(dy, dy_righttype)
     end
     return y, ty
 end

DifferentiationInterface/ext/DifferentiationInterfaceMooncakeExt/onearg.jl

@@ -79,7 +79,7 @@ function DI.value_and_pullback!(
     ) where {F, C}
     DI.check_prep(f, prep, backend, x, ty, contexts...)
     y, new_tx = DI.value_and_pullback(f, prep, backend, x, ty, contexts...)
-    foreach(copyto!, tx, new_tx)
+    foreach(copy!, tx, new_tx)
     return y, tx
 end
 
@@ -162,7 +162,7 @@ function DI.value_and_gradient!(
         prep.cache, f, x, map(DI.unwrap, contexts)...;
         prep.args_to_zero
     )
-    copyto!(grad, new_grad)
+    copy!(grad, new_grad)
     return y, grad
 end
 

DifferentiationInterface/ext/DifferentiationInterfaceMooncakeExt/twoarg.jl

@@ -57,7 +57,7 @@ function DI.value_and_pullback(
     DI.check_prep(f!, y, prep, backend, x, ty, contexts...)
     dy = only(ty)
     # Prepare cotangent to add after the forward pass.
-    dy_righttype_after = copyto!(prep.dy_righttype, dy)
+    dy_righttype_after = copy!(prep.dy_righttype, dy)
     # Run the reverse-pass and return the results.
     y_after, (_, _, _, dx) = value_and_pullback!!(
         prep.cache,
@@ -69,7 +69,7 @@ function DI.value_and_pullback(
         map(DI.unwrap, contexts)...;
         prep.args_to_zero
     )
-    copyto!(y, y_after)
+    copy!(y, y_after)
     return y, (_copy_output(dx),)
 end
 
@@ -84,7 +84,7 @@ function DI.value_and_pullback(
     ) where {F, C}
     DI.check_prep(f!, y, prep, backend, x, ty, contexts...)
     tx = map(ty) do dy
-        dy_righttype_after = copyto!(prep.dy_righttype, dy)
+        dy_righttype_after = copy!(prep.dy_righttype, dy)
         y_after, (_, _, _, dx) = value_and_pullback!!(
             prep.cache,
             dy_righttype_after,
@@ -95,7 +95,7 @@ function DI.value_and_pullback(
             map(DI.unwrap, contexts)...;
             prep.args_to_zero
         )
-        copyto!(y, y_after)
+        copy!(y, y_after)
         _copy_output(dx)
     end
     return y, tx
@@ -113,7 +113,7 @@ function DI.value_and_pullback!(
     ) where {F, C}
     DI.check_prep(f!, y, prep, backend, x, ty, contexts...)
     _, new_tx = DI.value_and_pullback(f!, y, prep, backend, x, ty, contexts...)
-    foreach(copyto!, tx, new_tx)
+    foreach(copy!, tx, new_tx)
     return y, tx
 end
 

DifferentiationInterface/ext/DifferentiationInterfaceReverseDiffExt/onearg.jl

@@ -241,7 +241,7 @@ function DI.value_and_jacobian!(
         result = jacobian!(result, f, x, prep.config)
     end
     y = DR.value(result)
-    jac === DR.jacobian(result) || copyto!(jac, DR.jacobian(result))
+    jac === DR.jacobian(result) || copy!(jac, DR.jacobian(result))
     return y, jac
 end
 
@@ -302,7 +302,7 @@ function DI.value_and_jacobian!(
     result = DiffResult(y, (jac,))
     result = jacobian!(result, fc, x, prep.config)
     y = DR.value(result)
-    jac === DR.jacobian(result) || copyto!(jac, DR.jacobian(result))
+    jac === DR.jacobian(result) || copy!(jac, DR.jacobian(result))
     return y, jac
 end
 

DifferentiationInterface/ext/DifferentiationInterfaceSparseMatrixColoringsExt/hessian.jl

@@ -125,7 +125,7 @@ function DI.hessian!(
         )
 
         for b in eachindex(batched_results[a])
-            copyto!(
+            copy!(
                 view(compressed_matrix, :, 1 + ((a - 1) * B + (b - 1)) % N),
                 vec(batched_results[a][b]),
             )

DifferentiationInterface/ext/DifferentiationInterfaceSparseMatrixColoringsExt/jacobian.jl

@@ -316,7 +316,7 @@ function _sparse_jacobian_aux!(
         )
 
         for b in eachindex(batched_results[a])
-            copyto!(
+            copy!(
                 view(compressed_matrix, :, 1 + ((a - 1) * B + (b - 1)) % N),
                 vec(batched_results[a][b]),
             )
@@ -365,7 +365,7 @@ function _sparse_jacobian_aux!(
             if eltype(x) <: Complex
                 batched_results[a][b] .= conj.(batched_results[a][b])
             end
-            copyto!(
+            copy!(
                 view(compressed_matrix, 1 + ((a - 1) * B + (b - 1)) % N, :),
                 vec(batched_results[a][b]),
             )

DifferentiationInterface/ext/DifferentiationInterfaceSparseMatrixColoringsExt/jacobian_mixed.jl

@@ -245,7 +245,7 @@ function _sparse_jacobian_aux!(
         )
 
         for b in eachindex(batched_results_forward[a])
-            copyto!(
+            copy!(
                 view(compressed_matrix_forward, :, 1 + ((a - 1) * Bf + (b - 1)) % Nf),
                 vec(batched_results_forward[a][b]),
             )
@@ -267,7 +267,7 @@ function _sparse_jacobian_aux!(
             if eltype(x) <: Complex
                 batched_results_reverse[a][b] .= conj.(batched_results_reverse[a][b])
             end
-            copyto!(
+            copy!(
                 view(compressed_matrix_reverse, 1 + ((a - 1) * Br + (b - 1)) % Nr, :),
                 vec(batched_results_reverse[a][b]),
             )

DifferentiationInterface/ext/DifferentiationInterfaceTrackerExt/DifferentiationInterfaceTrackerExt.jl

@@ -132,7 +132,7 @@ function DI.value_and_gradient!(
     ) where {C}
     DI.check_prep(f, prep, backend, x, contexts...)
     y, new_grad = DI.value_and_gradient(f, prep, backend, x, contexts...)
-    return y, copyto!(grad, new_grad)
+    return y, copy!(grad, new_grad)
 end
 
 function DI.gradient!(
@@ -144,7 +144,7 @@ function DI.gradient!(
         contexts::Vararg{DI.GeneralizedConstant, C},
     ) where {C}
     DI.check_prep(f, prep, backend, x, contexts...)
-    return copyto!(grad, DI.gradient(f, prep, backend, x, contexts...))
+    return copy!(grad, DI.gradient(f, prep, backend, x, contexts...))
 end
 
 end