Keyword Argument Not Working for @rsubset

[Ethan-Russell]

I'm running into an issue with @rsubset not passing through the view=true keyword argument in the way that I would expect. Here's a MWE:

julia> df = DataFrame(x=rand(4))
4×1 DataFrame
 Row │ x
     │ Float64   
─────┼───────────
   1 │ 0.0970303
   2 │ 0.436719
   3 │ 0.165089
   4 │ 0.505097

julia> subset(df, :x=>ByRow(x->x>0.5), view=true)
1×1 SubDataFrame
 Row │ x        
     │ Float64  
─────┼──────────
   1 │ 0.505097

julia> DataFramesMeta.@rsubset(df, :x>0.5, view=true)
1×1 DataFrame
 Row │ x        
     │ Float64  
─────┼──────────
   1 │ 0.505097

You can see above that @rsubset returns a DataFrame rather than a SubDataFrame as I would expect

[pdeffebach]

Yes. This is expected behavior. Keyword arguments were added in #323 , but we haven't made a release yet.

Note, your code is actually just saying "keep all rows" in the 2nd argument, so it is doing something just not what you think its doing.

[Ethan-Russell]

Thanks for the quick response. Good to know - I will look out for the next release.

I'm not sure that I'm tracking - how is my code saying to keep all the rows in the second argument? Isn't it only keeping the rows in which x>0.5?

[pdeffebach]

Sorry, multiple arguments are "and"-ed.

So the first argument says "keep all rows in which :x > .5" and the second one says "keep all rows". The intersection of those two conditions is to just keep all rows with :x > .5.

[Ethan-Russell]

Ah, I see - so currently the view=true keyword argument is being parsed as an argument, which is keeping all rows?

[pdeffebach]

yes exactly.

[Ethan-Russell]

Thanks! Closing since this is intended behavior.