Use handwritten rules for zero and one

Project.toml

@@ -1,6 +1,6 @@
 name = "ChainRules"
 uuid = "082447d4-558c-5d27-93f4-14fc19e9eca2"
-version = "1.59.0"
+version = "1.59.1"
 
 [deps]
 Adapt = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"

src/rulesets/Base/base.jl

@@ -2,11 +2,30 @@
 # that also have FastMath versions.
 
 @scalar_rule copysign(y, x) (ifelse(signbit(x)!=signbit(y), -one(y), +one(y)), NoTangent())
-
-@scalar_rule one(x) ZeroTangent()
-@scalar_rule zero(x) ZeroTangent()
 @scalar_rule transpose(x) true
 
+# `zero`
+
+function frule((_, _), ::typeof(zero), x)
+    return (zero(x), ZeroTangent())
+end
+
+function rrule(::typeof(zero), x)
+    zero_pullback(_) = (NoTangent(), ZeroTangent())
+    return (zero(x), zero_pullback)
+end
+
+# `one`
+
+function frule((_, _), ::typeof(one), x)
+    return (one(x), ZeroTangent())
+end
+
+function rrule(::typeof(one), x)
+    one_pullback(_) = (NoTangent(), ZeroTangent())
+    return (one(x), one_pullback)
+end
+
 # `adjoint`
 
 frule((_, Δz), ::typeof(adjoint), z::Number) = (z', Δz')

test/rulesets/Base/base.jl

@@ -1,4 +1,14 @@
 @testset "base.jl" begin
+    @testset "zero/one" begin
+        for f in [zero, one]
+            for x in [1.0, 1.0im, [10.0+im 11.0-im; 12.0+2im 13.0-3im]]
+                test_frule(f, x)
+                test_rrule(f, x)
+            end
+        end
+        test_frule(zero, [1.0, 2.0, 3.0])
+        test_rrule(zero, [1.0, 2.0, 3.0])
+    end
     @testset "copysign" begin
         # don't go too close to zero as the numerics may jump over it yielding wrong results
         @testset "at $y" for y in (-1.1, 0.1, 100.0)