- unique(), mad(), and iqr() are no longer auto-vectorized

- Bugfix: `@ungroup()` now preserves row-ordering (and is faster)
- Bugfix: `slice_sample()` now throws an error if no `n` or `prop` keyword argument is provided
- Bump minimum Julia version to 1.9

Author: kdpsingh

NEWS.md

@@ -1,5 +1,11 @@
 # TidierData.jl updates
 
+## v0.16.0 - 2024-06-07
+- `unique()`, `mad()`, and `iqr()` are no longer auto-vectorized
+- Bugfix: `@ungroup()` now preserves row-ordering (and is faster)
+- Bugfix: `slice_sample()` now throws an error if no `n` or `prop` keyword argument is provided
+- Bump minimum Julia version to 1.9
+
 ## v0.15.2 - 2024-04-19
 - Update Chain.jl dependency version
 

Project.toml

@@ -1,7 +1,7 @@
 name = "TidierData"
 uuid = "fe2206b3-d496-4ee9-a338-6a095c4ece80"
 authors = ["Karandeep Singh"]
-version = "0.15.2"
+version = "0.16.0"
 
 [deps]
 Chain = "8be319e6-bccf-4806-a6f7-6fae938471bc"
@@ -22,7 +22,7 @@ Reexport = "0.2, 1"
 ShiftedArrays = "2"
 Statistics = "1.6"
 StatsBase = "0.34, 1"
-julia = "1.6"
+julia = "1.9"
 
 [extras]
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"

src/TidierData.jl

@@ -28,7 +28,7 @@ const code = Ref{Bool}(false) # output DataFrames.jl code?
 const log = Ref{Bool}(false) # output tidylog output? (not yet implemented)
 
 # The global do-not-vectorize "list"
-const not_vectorized = Ref{Vector{Symbol}}([:getindex, :rand, :esc, :Ref, :Set, :Cols, :collect, :(:), :∘, :lag, :lead, :ntile, :repeat, :across, :desc, :mean, :std, :var, :median, :first, :last, :minimum, :maximum, :sum, :length, :skipmissing, :quantile, :passmissing, :cumsum, :cumprod, :accumulate, :is_float, :is_integer, :is_string, :cat_rev, :cat_relevel, :cat_infreq, :cat_lump, :cat_reorder, :cat_collapse, :cat_lump_min, :cat_lump_prop, :categorical, :as_categorical, :is_categorical])
+const not_vectorized = Ref{Vector{Symbol}}([:getindex, :rand, :esc, :Ref, :Set, :Cols, :collect, :(:), :∘, :lag, :lead, :ntile, :repeat, :across, :desc, :mean, :std, :var, :median, :mad, :first, :last, :minimum, :maximum, :sum, :length, :skipmissing, :quantile, :passmissing, :cumsum, :cumprod, :accumulate, :is_float, :is_integer, :is_string, :cat_rev, :cat_relevel, :cat_infreq, :cat_lump, :cat_reorder, :cat_collapse, :cat_lump_min, :cat_lump_prop, :categorical, :as_categorical, :is_categorical, :unique, :iqr])
 
 # The global do-not-escape "list"
 # `in`, `∈`, and `∉` should be vectorized in auto-vec but not escaped
@@ -494,7 +494,7 @@ end
 $docstring_ungroup
 """
 macro ungroup(df)
-  :(DataFrame($(esc(df))))
+  :(transform($(esc(df)); ungroup = true))
 end
 
 """
@@ -542,7 +542,7 @@ macro distinct(df, exprs...)
       # because if the original DataFrame is grouped, it must be ungrouped
       # and then regrouped, so there's no need to make a copy up front.
       # This is because `unique()` does not work on GroupDataFrames.
-      local df_copy = DataFrame($(esc(df)))
+      local df_copy = transform($(esc(df)); ungroup = true)
       if $any_found_n
         transform!(df_copy, nrow => :TidierData_n)
       end

src/slice.jl

@@ -64,7 +64,7 @@ macro slice_sample(df, exprs...)
                 as_integer(floor(n() * $expr_dict[:prop]));
                 replace=$replace))
     else
-      @slice($(esc(df)), sample(1:n(), 1; replace=$replace))
+      throw("Please provide either an `n` or a `prop` value as a keyword argument.")
     end
   end