Fix handling of variable_eltype in stack (#3043)

Project.toml

@@ -1,6 +1,6 @@
 name = "DataFrames"
 uuid = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
-version = "1.3.3"
+version = "1.3.4"
 
 [deps]
 Compat = "34da2185-b29b-5c13-b0c7-acf172513d20"
@@ -29,7 +29,7 @@ DataAPI = "1.9"
 InvertedIndices = "1"
 IteratorInterfaceExtensions = "0.1.1, 1"
 Missings = "0.4.2, 1"
-PooledArrays = "1.3.0"
+PooledArrays = "1.4.2"
 PrettyTables = "0.12, 1"
 Reexport = "0.1, 0.2, 1"
 ShiftedArrays = "1"

src/abstractdataframe/reshape.jl

@@ -164,7 +164,7 @@ function stack(df::AbstractDataFrame,
         # (note that copyto! inserts levels in their order of appearance)
         nms = names(df, ints_measure_vars)
         simnms = similar(nms, variable_eltype)
-        catnms = simnms isa Vector ? PooledArray(catnms) : simnms
+        catnms = simnms isa Vector ? PooledArray(simnms) : simnms
         copyto!(catnms, nms)
     end
     return DataFrame(AbstractVector[[repeat(df[!, c], outer=N) for c in ints_id_vars]..., # id_var columns

test/reshape.jl

@@ -730,4 +730,18 @@ end
     @test dfu.a isa Vector{Any}
 end
 
+@testset "variable_eltype in stack tests" begin
+    df = DataFrame(A=1:3, B=[2.0, -1.1, 2.8], C=["p", "q", "r"])
+    @test_throws MethodError stack(df, :C, variable_name=:D, variable_eltype=Int)
+    for T in (AbstractString, Any)
+        sdf = stack(df, [:A, :B], variable_name=:D, variable_eltype=T)
+        @test sdf == DataFrame(C=["p", "q", "r", "p", "q", "r"],
+                               D=["A", "A", "A", "B", "B", "B"],
+                               value=[1.0, 2.0, 3.0, 2.0, -1.1, 2.8])
+        @test sdf.C isa Vector{String}
+        @test sdf.value isa Vector{Float64}
+        @test sdf.D isa PooledVector{T}
+    end
+end
+
 end # module