question: how to handle `parsefield(io, Char)` about csv.jl HOT 3 CLOSED

Just added this. Note that columns aren't going to be detected automatically as Char, but you can specify a column to be Char type and it will use the new parsing function. Let me know if you run into any issues.

(also, you'll need current master for this to work obviously, just do Pkg.checkout("CSV"))

from csv.jl.

Thanks. I also found that I can handle encodings like this as

@enum Gender male=1 female=2

function Base.parse(::Type{Gender}, string)
    s = strip(string)
    if (s == "M")
        male
    elseif (s == "F")
        female
    else
        throw(ArgumentError("string $(s) not recognized as Gender"))
    end
end

CSV.parsefield(source, Gender)

Once other than Int32 Enums are supported, this should be really compact.

from csv.jl.

FWIW, CategoricalArrays should be easier to set up and as efficient for most cases. You can already use a UInt8 storage to save space.

from csv.jl.