Get majority of tests working with RealQuantity

Author: MilesCranmer

ext/DynamicQuantitiesUnitfulExt.jl

@@ -1,6 +1,7 @@
 module DynamicQuantitiesUnitfulExt
 
-import DynamicQuantities
+using DynamicQuantities: DynamicQuantities, ABSTRACT_QUANTITY_TYPES
+
 import Unitful
 import Unitful: @u_str
 
@@ -23,30 +24,33 @@ function unitful_equivalences()
     return NamedTuple((k => si_units[k] for k in keys(si_units)))
 end
 
-Base.convert(::Type{Unitful.Quantity}, x::DynamicQuantities.Quantity) =
-    let
-        validate_upreferred()
-        cumulator = DynamicQuantities.ustrip(x)
-        dims = DynamicQuantities.dimension(x)
-        if dims isa DynamicQuantities.SymbolicDimensions
-            throw(ArgumentError("Conversion of a `DynamicQuantities.Quantity` to a `Unitful.Quantity` is not defined with dimensions of type `SymbolicDimensions`. Instead, you can first use the `uexpand` function to convert the dimensions to their base SI form of type `Dimensions`, then convert this quantity to a `Unitful.Quantity`."))
+for (_, _, Q) in ABSTRACT_QUANTITY_TYPES
+    @eval begin
+        function Base.convert(::Type{Unitful.Quantity}, x::$Q)
+            validate_upreferred()
+            cumulator = DynamicQuantities.ustrip(x)
+            dims = DynamicQuantities.dimension(x)
+            if dims isa DynamicQuantities.SymbolicDimensions
+                throw(ArgumentError("Conversion of a `DynamicQuantities." * string($Q) * "` to a `Unitful.Quantity` is not defined with dimensions of type `SymbolicDimensions`. Instead, you can first use the `uexpand` function to convert the dimensions to their base SI form of type `Dimensions`, then convert this quantity to a `Unitful.Quantity`."))
+            end
+            equiv = unitful_equivalences()
+            for dim in keys(dims)
+                value = dims[dim]
+                iszero(value) && continue
+                cumulator *= equiv[dim]^value
+            end
+            cumulator
         end
-        equiv = unitful_equivalences()
-        for dim in keys(dims)
-            value = dims[dim]
-            iszero(value) && continue
-            cumulator *= equiv[dim]^value
+        function Base.convert(::Type{$Q}, x::Unitful.Quantity{T}) where {T}
+            return convert($Q{T,DynamicQuantities.DEFAULT_DIM_TYPE}, x)
+        end
+        function Base.convert(::Type{$Q{T,D}}, x::Unitful.Quantity) where {T,R,D<:DynamicQuantities.AbstractDimensions{R}}
+            value = Unitful.ustrip(Unitful.upreferred(x))
+            dimension = convert(D, Unitful.dimension(x))
+            return $Q(convert(T, value), dimension)
         end
-        cumulator
-    end
-
-Base.convert(::Type{DynamicQuantities.Quantity}, x::Unitful.Quantity{T}) where {T} = convert(DynamicQuantities.Quantity{T,DynamicQuantities.DEFAULT_DIM_TYPE}, x)
-Base.convert(::Type{DynamicQuantities.Quantity{T,D}}, x::Unitful.Quantity) where {T,R,D<:DynamicQuantities.AbstractDimensions{R}} =
-    let
-        value = Unitful.ustrip(Unitful.upreferred(x))
-        dimension = convert(D, Unitful.dimension(x))
-        return DynamicQuantities.Quantity(convert(T, value), dimension)
     end
+end
 
 Base.convert(::Type{DynamicQuantities.Dimensions}, d::Unitful.Dimensions) = convert(DynamicQuantities.DEFAULT_DIM_TYPE, d)
 Base.convert(::Type{DynamicQuantities.Dimensions{R}}, d::Unitful.Dimensions{D}) where {R,D} =

src/fixed_rational.jl

@@ -14,21 +14,23 @@ struct FixedRational{T<:Integer,den} <: Real
     num::T
     global unsafe_fixed_rational(num::Integer, ::Type{T}, ::Val{den}) where {T,den} = new{T,den}(num)
 end
+@inline _denom(::Type{F}) where {T,den,F<:FixedRational{T,den}} = den
 
 """
     denom(F::FixedRational)
 
 Since `den` can be a different type than `T`, this function
 is used to get the denominator as a `T`.
 """
-denom(::Type{F}) where {T,den,F<:FixedRational{T,den}} = convert(T, den)
+denom(::Type{<:F}) where {T,F<:FixedRational{T}} = convert(T, _denom(F))
 denom(x::FixedRational) = denom(typeof(x))
 
 # But, for Val(den), we need to use the same type as at init.
 # Otherwise, we would have type instability.
-val_denom(::Type{F}) where {T,den,F<:FixedRational{T,den}} = Val(den)
+val_denom(::Type{<:F}) where {F<:FixedRational} = Val(_denom(F))
 
-num_type(::Type{F}) where {T,F<:FixedRational{T}} = T
+num_type(::Type{<:FixedRational{T}}) where {T} = T
+num_type(x::FixedRational) = num_type(typeof(x))
 
 const DEFAULT_NUMERATOR_TYPE = Int32
 const DEFAULT_DENOM = DEFAULT_NUMERATOR_TYPE(2^4 * 3^2 * 5^2 * 7)
@@ -75,9 +77,9 @@ Base.decompose(x::F) where {T,F<:FixedRational{T}} = (x.num, zero(T), denom(F))
 
 # Promotion with self or rational-like
 function Base.promote_rule(::Type{F1}, ::Type{F2}) where {F1<:FixedRational,F2<:FixedRational}
-    denom(F1) == denom(F2) ||
+    _denom(F1) == _denom(F2) ||
         error("Refusing to promote `FixedRational` types with mixed denominators. Use `Rational` instead.")
-    return FixedRational{promote_type(num_type(F1), num_type(F2)),denom(F1)}
+    return FixedRational{promote_type(num_type(F1), num_type(F2)), _denom(F1)}
 end
 function Base.promote_rule(::Type{F}, ::Type{Rational{T2}}) where {F<:FixedRational,T2}
     return Rational{promote_type(num_type(F),T2)}

test/test_unitful.jl

@@ -6,10 +6,10 @@ import Ratios: SimpleRatio
 import SaferIntegers: SafeInt16
 using Test
 
-risapprox(x::Unitful.Quantity, y::Unitful.Quantity; kws...) =
-    let (xfloat, yfloat) = (Unitful.ustrip ∘ Unitful.upreferred).((x, y))
-        return isapprox(xfloat, yfloat; kws...)
-    end
+function risapprox(x::Unitful.Quantity, y::Unitful.Quantity; kws...)
+    (xfloat, yfloat) = (Unitful.ustrip ∘ Unitful.upreferred).((x, y))
+    return isapprox(xfloat, yfloat; kws...)
+end
 
 for T in [DEFAULT_VALUE_TYPE, Float16, Float32, Float64], R in [DEFAULT_DIM_BASE_TYPE, Rational{Int16}, Rational{Int32}, SimpleRatio{Int}, SimpleRatio{SafeInt16}]
     D = DynamicQuantities.Dimensions{R}

test/unittests.jl

@@ -1,8 +1,8 @@
 using DynamicQuantities
 using DynamicQuantities: FixedRational
-using DynamicQuantities: DEFAULT_DIM_BASE_TYPE, DEFAULT_DIM_TYPE, DEFAULT_VALUE_TYPE
+using DynamicQuantities: DEFAULT_QUANTITY_TYPE, DEFAULT_DIM_BASE_TYPE, DEFAULT_DIM_TYPE, DEFAULT_VALUE_TYPE
 using DynamicQuantities: array_type, value_type, dim_type, quantity_type
-using DynamicQuantities: GenericQuantity
+using DynamicQuantities: GenericQuantity, with_type_parameters, constructorof
 using Ratios: SimpleRatio
 using SaferIntegers: SafeInt16
 using StaticArrays: SArray, MArray
@@ -20,7 +20,7 @@ end
 
 @testset "Basic utilities" begin
 
-    for Q in [Quantity, GenericQuantity], T in [DEFAULT_VALUE_TYPE, Float16, Float32, Float64], R in [DEFAULT_DIM_BASE_TYPE, Rational{Int16}, Rational{Int32}, SimpleRatio{Int}, SimpleRatio{SafeInt16}]
+    for Q in [Quantity, GenericQuantity, RealQuantity], T in [DEFAULT_VALUE_TYPE, Float16, Float32, Float64], R in [DEFAULT_DIM_BASE_TYPE, Rational{Int16}, Rational{Int32}, SimpleRatio{Int}, SimpleRatio{SafeInt16}]
         D = Dimensions{R}
         x = Q(T(0.2), D, length=1, mass=2.5)
 
@@ -390,32 +390,32 @@ end
     @test utime(x) == -2
 
     y = 0.9u"sqrt(mΩ)"
-    @test typeof(y) == Quantity{Float64,DEFAULT_DIM_TYPE}
+    @test typeof(y) == with_type_parameters(DEFAULT_QUANTITY_TYPE, Float64, DEFAULT_DIM_TYPE)
     @test ustrip(y) ≈ 0.02846049894151541
     @test ucurrent(y) == -1
     @test ulength(y) == 1
 
     y = BigFloat(0.3) * u"mΩ"
-    @test typeof(y) == Quantity{BigFloat,DEFAULT_DIM_TYPE}
+    @test typeof(y) == with_type_parameters(DEFAULT_QUANTITY_TYPE, BigFloat, DEFAULT_DIM_TYPE)
     @test ustrip(y) ≈ 0.0003
     @test ulength(y) == 2
 
-    y32 = convert(Quantity{Float32,Dimensions{Rational{Int16}}}, y)
-    @test typeof(y32) == Quantity{Float32,Dimensions{Rational{Int16}}}
+    y32 = convert(with_type_parameters(DEFAULT_QUANTITY_TYPE, Float32, Dimensions{Rational{Int16}}), y)
+    @test typeof(y32) == with_type_parameters(DEFAULT_QUANTITY_TYPE, Float32, Dimensions{Rational{Int16}})
     @test ustrip(y32) ≈ 0.0003
 
     z = u"yr"
     @test utime(z) == 1
     @test ustrip(z) ≈ 60 * 60 * 24 * 365.25
 
     # Test type stability of extreme range of units
-    @test typeof(u"1") == Quantity{Float64,DEFAULT_DIM_TYPE}
-    @test typeof(u"1f0") == Quantity{Float64,DEFAULT_DIM_TYPE}
-    @test typeof(u"s"^2) == Quantity{Float64,DEFAULT_DIM_TYPE}
-    @test typeof(u"Ω") == Quantity{Float64,DEFAULT_DIM_TYPE}
-    @test typeof(u"Gyr") == Quantity{Float64,DEFAULT_DIM_TYPE}
-    @test typeof(u"fm") == Quantity{Float64,DEFAULT_DIM_TYPE}
-    @test typeof(u"fm"^2) == Quantity{Float64,DEFAULT_DIM_TYPE}
+    @test typeof(u"1") == DEFAULT_QUANTITY_TYPE
+    @test typeof(u"1f0") == DEFAULT_QUANTITY_TYPE
+    @test typeof(u"s"^2) == DEFAULT_QUANTITY_TYPE
+    @test typeof(u"Ω") == DEFAULT_QUANTITY_TYPE
+    @test typeof(u"Gyr") == DEFAULT_QUANTITY_TYPE
+    @test typeof(u"fm") == DEFAULT_QUANTITY_TYPE
+    @test typeof(u"fm"^2) == DEFAULT_QUANTITY_TYPE
 
     @test_throws LoadError eval(:(u":x"))
 end
@@ -486,10 +486,10 @@ end
     a = 0.5u"km/s"
     b = MyNumber(0.5)
     ar = [a, b]
-    @test ar isa Vector{Number}
+    @test ar isa Vector{Real}
     @test a === ar[1]
     @test b === ar[2]
-    @test promote_type(MyNumber, typeof(a)) == Number
+    @test promote_type(MyNumber, typeof(a)) == Real
 
     # Explicit conversion so coverage can see it:
     D = DEFAULT_DIM_TYPE
@@ -609,7 +609,7 @@ end
 
     q = 1.5us"km/s"
     @test q == 1.5 * us"km" / us"s"
-    @test typeof(q) <: Quantity{Float64,<:SymbolicDimensions}
+    @test typeof(q) <: with_type_parameters(DEFAULT_QUANTITY_TYPE, Float64, SymbolicDimensions{DEFAULT_DIM_BASE_TYPE})
     @test string(dimension(q)) == "s⁻¹ km"
     @test uexpand(q) == 1.5u"km/s"
     @test string(dimension(us"Constants.au^1.5")) == "au³ᐟ²"
@@ -683,13 +683,13 @@ end
     @test_throws DimensionError uconvert(us"nm * J", 5e-9u"m")
 
     # Types:
-    @test typeof(uconvert(us"nm", 5e-9u"m")) <: Quantity{Float64,<:SymbolicDimensions}
+    @test typeof(uconvert(us"nm", 5e-9u"m")) <: RealQuantity{Float64,<:SymbolicDimensions}
     @test typeof(uconvert(us"nm", GenericQuantity(5e-9u"m"))) <: GenericQuantity{Float64,<:SymbolicDimensions}
     @test uconvert(GenericQuantity(us"nm"), GenericQuantity(5e-9u"m")) ≈ 5us"nm"
     @test uconvert(GenericQuantity(us"nm"), GenericQuantity(5e-9u"m")) ≈ GenericQuantity(5us"nm")
 
     # We only want to convert the dimensions, and ignore the quantity type:
-    @test typeof(uconvert(GenericQuantity(us"nm"), 5e-9u"m")) <: Quantity{Float64,<:SymbolicDimensions}
+    @test typeof(uconvert(GenericQuantity(us"nm"), 5e-9u"m")) <: RealQuantity{Float64,<:SymbolicDimensions}
 
     q = 1.5u"Constants.M_sun"
     qs = uconvert(us"Constants.M_sun", 5.0 * q)
@@ -704,7 +704,7 @@ end
     VERSION >= v"1.8" &&
         @test_throws "You passed a quantity" uconvert(1.2us"m", 1.0u"m")
 
-    for Q in (Quantity, GenericQuantity)
+    for Q in (RealQuantity, Quantity, GenericQuantity)
         # Different types require converting both arguments:
         q = convert(Q{Float16}, 1.5u"g")
         qs = uconvert(convert(Q{Float16}, us"g"), 5 * q)
@@ -769,10 +769,10 @@ end
     x = 1.0u"m"
     y = x ^ (3//2)
     @test y == Quantity(1.0, length=3//2)
-    @test typeof(y) == Quantity{Float64,DEFAULT_DIM_TYPE}
+    @test typeof(y) == RealQuantity{Float64,DEFAULT_DIM_TYPE}
 end
 
-for Q in (Quantity, GenericQuantity)
+for Q in (RealQuantity, Quantity, GenericQuantity)
     @testset "Arrays" begin
         @testset "Basics" begin
             x = QuantityArray(randn(32), Q(u"km/s"))
@@ -798,7 +798,7 @@ for Q in (Quantity, GenericQuantity)
 
             # Test default constructors:
             @test QuantityArray(ones(3), u"m/s") == QuantityArray(ones(3), length=1, time=-1)
-            @test typeof(QuantityArray(ones(3), u"m/s")) <: QuantityArray{Float64,1,<:Dimensions,<:Quantity,<:Array}
+            @test typeof(QuantityArray(ones(3), u"m/s")) <: QuantityArray{Float64,1,<:Dimensions,<:constructorof(DEFAULT_QUANTITY_TYPE),<:Array}
 
             # We can create quantity arrays with generic quantity
             @test typeof(QuantityArray([[1.0], [2.0, 3.0]], dimension(u"m/s"))) <: QuantityArray{<:Any,1,<:Dimensions,<:GenericQuantity,<:Array}
@@ -964,14 +964,14 @@ for Q in (Quantity, GenericQuantity)
             @test ustrip(x .* y) == ustrip(x) .* ustrip(y)
         end
 
-        Q == Quantity && @testset "Broadcast different arrays" begin
+        Q in (Quantity, RealQuantity) && @testset "Broadcast different arrays" begin
             f(x, y, z, w) = x * y + z * w
             g(x, y, z, w) = f.(x, y, z, w)
 
             x = randn(32)
             y = QuantityArray(randn(32), u"km/s")
             z = rand(1:10, 32)
-            w = Quantity{Float32}(u"m/s")
+            w = Q{Float32}(u"m/s")
             @test typeof(g(x, y, z, w)) <: QuantityArray{Float64}
 
             y32 = QuantityArray(ustrip(y), dimension(y))
@@ -990,7 +990,7 @@ for Q in (Quantity, GenericQuantity)
             @test typeof(b .* y) <: QuantityArray{Float64}
         end
 
-        Q == Quantity && @testset "Broadcast scalars" begin
+        Q in (RealQuantity, Quantity) && @testset "Broadcast scalars" begin
             for (x, qx) in ((0.5, 0.5u"s"), ([0.5, 0.2], GenericQuantity([0.5, 0.2], time=1)))
                 @test size(qx) == size(x)
                 @test length(qx) == length(x)
@@ -1128,8 +1128,8 @@ end
         qy = QuantityArray(y; length=1)
 
         @test typeof(convert(typeof(qx), qy)) == typeof(qx)
-        @test convert(typeof(qx), qy)[1] isa Quantity{Float64}
-        @test convert(typeof(qx), qy)[1] == convert(Quantity{Float64}, qy[1])
+        @test convert(typeof(qx), qy)[1] isa RealQuantity{Float64}
+        @test convert(typeof(qx), qy)[1] == convert(RealQuantity{Float64}, qy[1])
     end
 end
 
@@ -1152,21 +1152,21 @@ end
         :log, :log2, :log10, :log1p, :exp, :exp2, :exp10, :expm1, :frexp, :exponent,
         :atan, :atand
     )
-    for Q in (Quantity, GenericQuantity), D in (Dimensions, SymbolicDimensions), f in functions
+    for Q in (RealQuantity, Quantity, GenericQuantity), D in (Dimensions, SymbolicDimensions), f in functions
         # Only test on valid domain
         valid_inputs = filter(
             x -> is_input_valid(eval(f), x),
             5rand(100) .- 2.5
         )
         for x in valid_inputs[1:3]
-            qx_dimensionless = Quantity(x, D)
-            qx_dimensions = Quantity(x, convert(D, dimension(u"m/s")))
+            qx_dimensionless = Q(x, D)
+            qx_dimensions = Q(x, convert(D, dimension(u"m/s")))
             @eval @test $f($qx_dimensionless) == $f($x)
             @eval @test_throws DimensionError $f($qx_dimensions)
             if f in (:atan, :atand)
                 for y in valid_inputs[end-3:end]
-                    qy_dimensionless = Quantity(y, D)
-                    qy_dimensions = Quantity(y, convert(D, dimension(u"m/s")))
+                    qy_dimensionless = Q(y, D)
+                    qy_dimensions = Q(y, convert(D, dimension(u"m/s")))
                     @eval @test $f($y, $qx_dimensionless) == $f($y, $x)
                     @eval @test $f($qy_dimensionless, $x) == $f($y, $x)
                     @eval @test $f($qy_dimensionless, $qx_dimensionless) == $f($y, $x)
@@ -1189,8 +1189,9 @@ end
         :floor, :trunc, :ceil, :significand,
         :ldexp, :round,
     )
-    for Q in (Quantity, GenericQuantity), D in (Dimensions, SymbolicDimensions), f in functions
+    for Q in (RealQuantity, Quantity, GenericQuantity), D in (Dimensions, SymbolicDimensions), f in functions
         T = f in (:abs, :real, :imag, :conj) ? ComplexF64 : Float64
+        T <: Complex && Q == RealQuantity && continue
         if f == :modf  # Functions that return multiple outputs
             for x in 5rand(T, 3) .- 2.5
                 dim = convert(D, dimension(u"m/s"))
@@ -1251,7 +1252,7 @@ end
 end
 
 @testset "Test div" begin
-    for Q in (Quantity, GenericQuantity)
+    for Q in (RealQuantity, Quantity, GenericQuantity)
         x = Q{Int}(10, length=1)
         y = Q{Int}(3, mass=-1)
         @test div(x, y) == Q{Int}(3, length=1, mass=1)