Make One, Zero, and DNE inferable when broadcasting with StaticArrays

Project.toml

@@ -1,17 +1,19 @@
 name = "ChainRulesCore"
 uuid = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"
-version = "0.5.2"
+version = "0.5.3"
 
 [deps]
 MuladdMacro = "46d2c3a1-f734-5fdb-9937-b9b9aeba4221"
 
 [compat]
-julia = "^1.0"
 MuladdMacro = "0.2.1"
+StaticArrays = "0.11, 0.12"
+julia = "^1.0"
 
 [extras]
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
+StaticArrays = "90137ffa-7385-5640-81b9-e52037218182"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [targets]
-test = ["Test", "LinearAlgebra"]
+test = ["Test", "LinearAlgebra", "StaticArrays"]

src/differentials/does_not_exist.jl

@@ -33,6 +33,7 @@ function extern(x::DoesNotExist)
 end
 
 Base.Broadcast.broadcastable(::DoesNotExist) = Ref(DoesNotExist())
+Base.Broadcast.broadcasted(::Type{DoesNotExist}) = DoesNotExist()
 
 Base.iterate(x::DoesNotExist) = (x, nothing)
 Base.iterate(::DoesNotExist, ::Any) = nothing

src/differentials/one.jl

@@ -8,6 +8,7 @@ struct One <: AbstractDifferential end
 extern(x::One) = true  # true is a strong 1.
 
 Base.Broadcast.broadcastable(::One) = Ref(One())
+Base.Broadcast.broadcasted(::Type{One}) = One()
 
 Base.iterate(x::One) = (x, nothing)
 Base.iterate(::One, ::Any) = nothing

src/differentials/zero.jl

@@ -8,6 +8,7 @@ struct Zero <: AbstractDifferential end
 extern(x::Zero) = false  # false is a strong 0. E.g. `false * NaN = 0.0`
 
 Base.Broadcast.broadcastable(::Zero) = Ref(Zero())
+Base.Broadcast.broadcasted(::Type{Zero}) = Zero()
 
 Base.iterate(x::Zero) = (x, nothing)
 Base.iterate(::Zero, ::Any) = nothing

test/rules.jl

@@ -1,5 +1,6 @@
 #######
 # Demo setup
+using StaticArrays: @SVector
 
 cool(x) = x + 1
 cool(x, y) = x + y + 1
@@ -11,6 +12,9 @@ dummy_identity(x) = x
 nice(x) = 1
 @scalar_rule(nice(x), Zero())
 
+very_nice(x, y) = x + y
+@scalar_rule(very_nice(x, y), (One(), One()))
+
 #######
 
 _second(t) = Base.tuple_type_head(Base.tuple_type_tail(t))
@@ -46,4 +50,9 @@ _second(t) = Base.tuple_type_head(Base.tuple_type_tail(t))
     @test nice_pushforward === 0
     rrx, nice_pullback = rrule(nice, 1)
     @test (NO_FIELDS, 0) === nice_pullback(1)
+
+    sx = @SVector [1, 2]
+    sy = @SVector [3, 4]
+    # This actually is testing that @scalar_rule and `One()` play nice together, w.r.t broadcasting
+    @inferred frule(very_nice, 1, 2, Zero(), sx, sy)
 end