# \[ANN\] DataFrameMacros.jl v0.3.0

**URL:** https://discourse.julialang.org/t/ann-dataframemacros-jl-v0-3-0/85771
**Category:** Package Announcements
**Tags:** macros, dataframes
**Created:** [August 15, 2022, 12:17pm UTC](https://discourse.julialang.org/t/ann-dataframemacros-jl-v0-3-0/85771 "2022-08-15T12:17:18Z")
**Posts on this page:** 1
**Page:** 1

<div class="post-metadata">

### Author: ![jules](https://avatars.discourse-cdn.com/v4/letter/j/41988e/32.png) [@jules](https://discourse.julialang.org/u/jules)
#### Post date: [August 15, 2022, 12:17pm UTC](https://discourse.julialang.org/t/ann-dataframemacros-jl-v0-3-0/85771/1 "2022-08-15T12:17:18Z")

</div>

I made some breaking changes to [DataFrameMacros](https://jkrumbiegel.com/DataFrameMacros.jl/dev/), which hopefully make the package easier to use and more powerful.

- 
  - **Breaking** : The `$()` interpolation syntax is replaced by `{}` for single columns (or broadcasted multi-columns)

- 
  - Added `{{}}` for referring to multiple columns as a tuple.

- 
  - **Breaking** : No more flag macros, replaced by explicit `@byrow`, `@bycol`, `@passmissing`, `@astable` which is also what DataFramesMeta uses and which I was finally convinced is better (less confusing) than the single-character macros I liked at first for their brevity.

- 
  - **Breaking** : `All()`, `Between()` and `Not()` have to be interpolated with `{}` and can’t be used standalone anymore.

An example of the new `{{}}` syntax is below. I was annoyed for a long time that it wasn’t as easy to refer to multiple columns together in one expression to aggregate over them and compare with some other columns at the same time. I already had implicit broadcasting for `{}` but that only makes it easier to run the same expression on multiple columns each, not multiple columns together. The `{{}}` syntax is replaced with a tuple of the specified columns, so you can run aggregations on that. Because of the tuple-ization it’s probably not useful for very large numbers of columns (too much compilation overhead I assume) but for normal workloads it should be fine.

```julia
julia> df = DataFrame(
           jan = randn(5),
           feb = randn(5),
           mar = randn(5),
           apr = randn(5),
           may = randn(5),
           jun = randn(5),
           jul = randn(5),
       )

```

```julia
5×7 DataFrame
 Row │ jan feb mar apr may jun jul ⋯
     │ Float64 Float64 Float64 Float64 Float64 Float64 Floa ⋯
─────┼──────────────────────────────────────────────────────────────────────────
   1 │ 1.19027 -0.664713 -0.339366 0.368002 -0.979539 1.52392 -0.8 ⋯
   2 │ 2.04818 0.980968 -0.843878 -0.281133 0.260402 -1.77773 0.3
   3 │ 1.14265 -0.0754831 -0.888936 -0.734886 -0.468489 -2.93306 -0.1
   4 │ 0.459416 0.273815 0.327215 -0.71741 -0.880897 0.782258 2.3
   5 │ -0.396679 -0.194229 0.592403 -0.77507 0.277726 2.31358 -0.9 ⋯
                                                                1 column omitted

```

```julia
julia> @select(df, :july_larger = :jul > median({{Between(:jan, :jun)}}))

```

```julia
5×1 DataFrame
 Row │ july_larger
     │ Bool
─────┼─────────────
   1 │ false
   2 │ true
   3 │ true
   4 │ true
   5 │ false

```

```julia
julia> @select(df, :mean_smaller = mean({{All()}}) < median({{All()}}))

```

```julia
5×1 DataFrame
 Row │ mean_smaller
     │ Bool
─────┼──────────────
   1 │ false
   2 │ true
   3 │ true
   4 │ false
   5 │ false

```