potential bug when using log in mutate

[rgreminger]

Hi, it seems like there might be a bug when using @mutate to create a new variable as the log of another one. Below is a simple example for such a case, and the error it produces in the latest release v0.16.5.

using TidierData

df = DataFrame(a = 'a':'e', b = 1.0:5.0);

df = @chain df begin
    @mutate(log_b = log(b)) 
end 

# Error it produces: 
ERROR: MethodError: objects of type Base.RefValue{Bool} are not callable
The object of type `Base.RefValue{Bool}` exists, but no method is defined for this combination of argument types when trying to treat it as a callable object.

[kdpsingh]

Thanks for reporting - will take a look!

[drizk1]

Looks like this is a temp work around, not sure what the best way to fix the underlying parsing function is, will defer to kdpsingh

julia> df = @chain df begin
           @mutate(log_b = Base.log(b)) 
       end
5×3 DataFrame
 Row │ a     b        log_b    
     │ Char  Float64  Float64  
─────┼─────────────────────────
   1 │ a         1.0  0.0
   2 │ b         2.0  0.693147
   3 │ c         3.0  1.09861
   4 │ d         4.0  1.38629
   5 │ e         5.0  1.60944

[kdpsingh]

This is definitely a bug. Will aim to fix shortly. Have been meaning to simplify the parsing engine which should resolve bugs of this sort.

[drizk1]

This should be fixed in v.17.1. I think i will leave this open for the time being tho. if any new edges come up, please let us know

julia> @chain df begin
                  @mutate(log_b = log(b)) 
              end
5×3 DataFrame
 Row │ a     b        log_b    
     │ Char  Float64  Float64  
─────┼─────────────────────────
   1 │ a         1.0  0.0
   2 │ b         2.0  0.693147
   3 │ c         3.0  1.09861
   4 │ d         4.0  1.38629
   5 │ e         5.0  1.60944