[WIP] Merge ComponentVectors

src/ComponentArrays.jl

@@ -2,6 +2,7 @@ module ComponentArrays
 
 import ChainRulesCore
 import StaticArrayInterface, ArrayInterface, Functors
+import Base.merge
 
 using LinearAlgebra
 

src/componentarray.jl

@@ -314,7 +314,7 @@ directly on an `AbstractAxis`.
 
 # Examples
 
-```julia-repl
+```jldoctest
 julia> using ComponentArrays
 
 julia> ca = ComponentArray(a=1, b=[1,2,3], c=(a=4,))
@@ -336,3 +336,44 @@ julia> sum(prod(ca[k]) for k in valkeys(ca))
     return :($k)
 end
 valkeys(ca::ComponentVector) = valkeys(getaxes(ca)[1])
+
+"""
+    merge(cvec1::ComponentVector, cvecs::ComponentVector...)
+
+Construct a new ComponentVector by merging two or more existing ones, in a left-associative
+manner. If a key is present in two or more CVectors, the right-most CVector takes priority.
+The type of the resulting CVector will be promoted to accomodate all the types of the merged
+CVectors
+
+# Examples
+```jldoctest
+julia> c1 = ComponentArray(a=1.2, b=2.3)
+ComponentVector{Float64}(a = 1.2, b = 2.3)
+
+julia> c2 = ComponentArray(a=1,h=4)
+ComponentVector{Int64}(a = 1, h = 4)
+
+julia> merge(c1,c2)
+ComponentVector{Float64}(a = 1.0, b = 2.3, h = 4.0)
+
+julia> merge(c2,c1)
+ComponentVector{Float64}(a = 1.2, h = 4.0, b = 2.3)
+```
+"""
+function Base.merge(ca::ComponentVector{T}, ca2::ComponentVector{T}) where T
+    ax = getaxes(ca)
+    ax2 = getaxes(ca2)
+    vks = valkeys(ax[1])
+    vks2 = valkeys(ax2[1])
+    _p = Vector{T}()
+    sizehint!(_p, length(c1) + length(ca2))
+    for vk in vks
+        if vk in vks2
+            _p = vcat(_p, ca2[vk])
+        else
+            _p = vcat(_p, ca[vk])
+        end
+    end
+    ComponentArray(_p, merged_ax)
+end
+Base.merge(a::ComponentVector, b::ComponentVector, cs::ComponentVector) = merge(merge(a,b), cs...)