`read_csv` needs DataFrames.jl which is a `require` import

[Dsantra92]

Reproduce the bug:

using MLDatasets
data = Iris()

StackTrace:

ERROR: Add `import DataFrames` or `using DataFrames` to your code to unlock this functionality.
Stacktrace:
 [1] error(s::String)
   @ Base ./error.jl:33
 [2] assert_imported
   @ /test-env/MLDatasets.jl/src/require.jl:30 [inlined]
 [3] require_import
   @ /test-env/MLDatasets.jl/src/require.jl:37 [inlined]
 [4] getproperty
   @ /test-env/MLDatasets.jl/src/require.jl:23 [inlined]
 [5] read_csv_asdf(path::String; kws::Base.Pairs{Symbol, Int64, Tuple{Symbol}, NamedTuple{(:header,), Tuple{Int64}}})
   @ MLDatasets /test-env/MLDatasets.jl/src/io.jl:16
 [6] #read_csv#9
   @ /test-env/MLDatasets.jl/src/io.jl:3 [inlined]
 [7] Iris(; dir::Nothing, as_df::Bool)
   @ MLDatasets /test-env/MLDatasets.jl/src/datasets/misc/iris.jl:82
 [8] Iris()
   @ MLDatasets /test-env/MLDatasets.jl/src/datasets/misc/iris.jl:81
 [9] top-level scope
   @ REPL[3]:1

Ps: Tested on my system and docker image 1.7.2

[johnnychen94]

This is an intended behavior introduced in #128. As indicated by the error message, what you need to do is:

using MLDatasets
using DataFrames
data = Iris()

[Dsantra92]

Thanks, I should have checked the new docs.