UndefVarError when using `@eval` in user-defined function

[ForceBru]

Even though the documentation says that @eval works inside of user-defined functions, here's an example where it doesn't:

julia> import Pkg; Pkg.activate(temp=true); Pkg.add(name="TidierData", version="0.17.0")

julia> using TidierData

julia> df1 = DataFrame(trainsz=rand([200, 140, 110, 80], 100), loss=rand(100), p=rand([1,2,3], 100));

julia> df2 = DataFrame(trainsz=rand([200, 140, 110, 80], 100), loss=rand(100), p=rand([1,2,3], 100));

julia> function plot_comparison(losses_ARp, what)
               for (i, trainsz) in enumerate([200, 140, 110, 80])
                       df_ARp = @eval @filter(losses_ARp, trainsz == $trainsz)
                       df_ARp_comp = @eval @filter(what, trainsz == $trainsz)
               end
       end;

julia> plot_comparison(df1, df2)
ERROR: UndefVarError: `losses_ARp` not defined in `Main`
Suggestion: check for spelling errors or missing imports.
Stacktrace:
 [1] macro expansion
   @ ~/.julia/packages/TidierData/59JUA/src/TidierData.jl:422 [inlined]
 [2] top-level scope
   @ REPL[5]:415
 [3] eval
   @ ./boot.jl:430 [inlined]
 [4] plot_comparison(losses_ARp::DataFrame, what::DataFrame)
   @ Main ./REPL[5]:3
 [5] top-level scope
   @ REPL[6]:1

Are there any workarounds?

[drizk1]

I think you might need to add $ for the function variables

function plot_comparison(losses_ARp, what)
               for (i, trainsz) in enumerate([200, 140, 110, 80])
                       df_ARp = @eval @filter($losses_ARp, trainsz == $trainsz)
                       df_ARp_comp = @eval @filter($what, trainsz == $trainsz)
               end
       end;

This does not error for me.

[ForceBru]

Thanks, @drizk1, this seems to work. I guess the rule is "when using @eval, interpolate all local variables like $variable".