Unexpected output from @combine when used in Lazy:@>

[dbuffoni]

Hi,
Apparently @combine seems to provide an unexpected result when used in a @> pipe after a transform.

Please review the following test:

`using DataFrames, DataFramesMeta
using Lazy: @>

df = DataFrame(a = [1, 1, 2, 2], b = [1, 3, 8, 10])
@> begin
groupby(df, :a)
#@Transform(c = :b.+1)
@combine(m=sum(:b))
end`

which correctly returns

2×2 DataFrame │ Row │ a │ m │ │ 1 │ 1 │ 4 │ │ 2 │ 2 │ 18 │

while uncommenting the @Transform the groups are lost and the entire block returns a single row

1×1 DataFrame │ Row │ m │ │ 1 │ 22 │

Is this expected? the workaround would be using 'by' instead of 'combine'

[pdeffebach]

Thanks for the issue report!

This is indeed expected. By default DataFrames.transform "ungroups" he data after the transformation. Fixing #179 to allow keyword arguments is the solution. Eventually it would be nice to do

@transform(m = sum(:b); ungroup = false)

which would allow grouping to be preserved in the pipe.

But this has no been added yet, unfortunately.

[dbuffoni]

Great, thank you for your explanation