weird typemap behavior (bug?)

[kragol]

The typemap option to CSV.File produces incorrect types when mapping Integer types (I haven't really tested other types though). Basically, all Int columns are converted to Float64 whatever the typemap indicates as a target type.

MWE:
file test.csv:

"a","b","c"
1,"a",12
1,"a",12

julia> CSV.File("test.csv"),typemap=Dict(Int64=>Int32))
2-element CSV.File{false}:
 CSV.Row: (a = 1.0, b = "a", c = 12.0)
 CSV.Row: (a = 1.0, b = "a", c = 12.0)

Here, the values of a and c are converted to Float64 instead of Int32 as requested.

julia 1.4.2
CSV 0.7.2

[quinnj]

Hmmm, yes, this is a bug. It turns out typemap wasn't adjusted property for our new "custom types" life we enjoy now. It still assumes you will only map types between the "standard" types (Int64, Float64, Bool, Date, DateTime, Time, Bool). I'll start noodling on how to make sure we're correct here.