Merge pull request #5 from ReactiveBayes/dev-v1.0.2

Fix ForwardDiff compatibility

Project.toml

@@ -1,13 +1,15 @@
 name = "TinyHugeNumbers"
 uuid = "783c9a47-75a3-44ac-a16b-f1ab7b3acf04"
 authors = ["Bagaev Dmitry <[email protected]> and contributors"]
-version = "1.0.1"
+version = "1.0.2"
 
 [compat]
 julia = "1"
 
 [extras]
+Aqua = "4c88cf16-eb10-579e-8560-4a9242c79595"
+ForwardDiff = "f6369f11-7733-5829-9624-2563aa707210"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [targets]
-test = ["Test"]
+test = ["Aqua", "ForwardDiff", "Test"]

src/TinyHugeNumbers.jl

@@ -120,8 +120,24 @@ const huge = HugeNumber()
 
 ## ------------------------------------------------------------------------------------ ##
 
-Base.promote_rule(::Type{Union{TinyNumber, HugeNumber}}, ::Type{T}) where {T} = T
-Base.promote_rule(::Type{TinyNumber}, ::Type{HugeNumber}) = Union{TinyNumber, HugeNumber}
-Base.promote_rule(::Type{HugeNumber}, ::Type{TinyNumber}) = Union{TinyNumber, HugeNumber}
+# A special structure that is used to promote `TinyNumber` and `HugeNumber` to the same type
+# but it cannot be instantiated, this might be useful in situations like `clamp(value, tiny, huge)`
+# in this case Julia attempts first to promote `tiny` and `huge` to the same type and then
+# uses the result to promote `value` to the resulting type. However, there is no "common" type for
+# both `tiny` and `huge` (except for the `Union` but we can't use it either since it introduces ambiguities) 
+# so we introduce a special structure that will accomodate that
+# see also: https://github.com/ReactiveBayes/TinyHugeNumbers.jl/issues/3
+# note: as a result, we cannot store `tiny` and `huge` in the same container (e.g. `Array`), 
+# but `[ 1.0, tiny, huge ]` will work just fine
+struct PromoteTinyOrHuge 
+    PromoteTinyOrHuge() = error("Cannot instantiate an internal structure for promotion.")
+end
+
+Base.promote_rule(::Type{T}, ::Type{PromoteTinyOrHuge}) where {T<:Real} = T
+Base.promote_rule(::Type{PromoteTinyOrHuge}, ::Type{PromoteTinyOrHuge}) = PromoteTinyOrHuge
+Base.promote_rule(::Type{TinyNumber}, ::Type{HugeNumber}) = PromoteTinyOrHuge
+
+Base.convert(::Type{PromoteTinyOrHuge}, ::TinyNumber) = error("Cannot convert `tiny` to `huge`. Are you trying to put `tiny` and `huge` in the same container (e.g. `Array`)?")
+Base.convert(::Type{PromoteTinyOrHuge}, ::HugeNumber) = error("Cannot convert `huge` to `tiny`. Are you trying to put `tiny` and `huge` in the same container (e.g. `Array`)?")
 
 end

test/runtests.jl

@@ -1,18 +1,19 @@
-using TinyHugeNumbers
-using Test
+using TinyHugeNumbers, Aqua, Test
+
+Aqua.test_all(TinyHugeNumbers, deps_compat=(; check_extras=false, check_weakdeps=true))
 
 import TinyHugeNumbers: TinyNumber, HugeNumber
 
-struct ArbitraryFloatType <: AbstractFloat end 
+struct ArbitraryFloatType <: AbstractFloat end
 
 @testset "TinyHugeNumbers.jl" begin
-
-    Base.eps(::Type{ ArbitraryFloatType }) = 0.1
-    Base.convert(::Type{ ArbitraryFloatType }, ::Integer) = ArbitraryFloatType() # for testing
+    
+    Base.eps(::Type{ArbitraryFloatType}) = 0.1
+    Base.convert(::Type{ArbitraryFloatType}, ::Integer) = ArbitraryFloatType() # for testing
 
     @test repr(tiny) == "tiny"
     @test repr(huge) == "huge"
-    
+
     @test typeof(tiny) === TinyNumber
     @test typeof(huge) === HugeNumber
 
@@ -77,7 +78,7 @@ struct ArbitraryFloatType <: AbstractFloat end
 
     for a in (1, 1.0, 0, 0.0, 1.0f0, 0.0f0, Int32(0), Int32(1), big"1", big"1.0", big"0", big"0.0")
         T = typeof(a)
-        for v in [tiny, huge]
+        for v in Real[tiny, huge]
             V = typeof(v)
 
             for op in [+, -, *, /, >, >=, <, <=]
@@ -109,3 +110,43 @@ struct ArbitraryFloatType <: AbstractFloat end
     end
 
 end
+
+@testset "ForwardDiff.jl compatibility" begin
+    import ForwardDiff
+
+    f(x) = clamp(x, tiny, huge)
+
+    @test @inferred(ForwardDiff.derivative(f, 1.0)) === 1.0
+    @test @inferred(ForwardDiff.derivative(f, 2.0)) === 1.0
+    @test @inferred(ForwardDiff.derivative(f, 0.0)) === 0.0
+    @test @inferred(ForwardDiff.derivative(f, huge + 1.0)) === 0.0
+    @test @inferred(ForwardDiff.derivative(f, tiny - 1.0)) === 0.0
+
+    g(x) = clamp(x^2, tiny, huge)
+
+    @test @inferred(ForwardDiff.derivative(g, 1.0)) === 2.0
+    @test @inferred(ForwardDiff.derivative(g, 2.0)) === 4.0
+    @test @inferred(ForwardDiff.derivative(g, 0.0)) === 0.0
+    @test @inferred(ForwardDiff.derivative(g, huge + 1.0)) === 0.0
+    @test @inferred(ForwardDiff.derivative(g, tiny - 1.0)) === 2(tiny - 1.0)
+end
+
+@testset "Storing `tiny` and `huge` in arrays" begin
+    @static if VERSION >= v"1.10"
+        @test_throws "Cannot convert `tiny` to `huge`" [tiny, huge]
+        @test_throws "Cannot convert `huge` to `tiny`" [huge, tiny]
+    else
+        @test_throws ErrorException [tiny, huge]
+        @test_throws ErrorException [huge, tiny]
+    end
+
+    for a in (1.0, 0.0, 1.0f0, 0.0f0, big"1.0", big"0.0")
+        @test [a, tiny, huge] == [a, tiny(a), huge(a)]
+        @test [tiny, a, huge] == [tiny(a), a, huge(a)]
+        @test [tiny, huge, a] == [tiny(a), huge(a), a]
+
+        @test [a, huge, tiny] == [a, huge(a), tiny(a)]
+        @test [huge, a, tiny] == [huge(a), a, tiny(a)]
+        @test [huge, tiny, a] == [huge(a), tiny(a), a]
+    end
+end