MariaDB version 11 is incompatible with MySQL - String columns are detected as Dec64

[hhaensel]

column types are not correctly inferred when updating mariadb to version 11. It's probably worthwhile switching to the latest mariadb c connector with MariaDB Server 10:

julia> DataFrame(DBInterface.execute(conn, "select * from `benutzer`"))
2×1 DataFrame
 Row │ uname
     │ String?
─────┼─────────
   1 │ test
   2 │ rest

with MariaDB Server 11:

julia> DBInterface.execute(conn, "select * from `benutzer`") |> DataFrame
ERROR: ArgumentError: invalid number format test
Stacktrace:
  [1] parse(#unused#::Type{DecFP.Dec64}, s::String)
    @ DecFP C:\Users\helmu\.julia\packages\DecFP\Cud5F\src\DecFP.jl:323

If I store strings that can be parsed as numbers in the field 'uname' the columns are indeed read as Dec64.

julia> DataFrame(hh)
3×1 DataFrame
 Row │ uname
     │ Dec64
─────┼───────
   1 │  10.0
   2 │  11.0
   3 │  14.0

This behaviour is consistent with the (wrong) column type stored in the query result:

julia> res = DBInterface.execute(conn, "select * from `benutzer`");

julia> res.types
1-element Vector{Type}:
 Union{Missing, DecFP.Dec64}

[pascalr0410]

Hello,

Same here, must fall back on branch 10 of MariaDb.

The C Lib used seems quite old, released in December. 2019.

In fact, after some debugging, the library returns a zero in field_type, causing it to be misinterpreted as Dec64

/src/MySQL.jl

# --> Here, the real type is VARCHAR

function juliatype(field_type, notnullable, isunsigned, isbinary, date_and_time)
    @info field_type # --> 0
    T = API.juliatype(field_type)
    @info T # --> Dec64
    T2 = isunsigned && !(T <: AbstractFloat) ? unsigned(T) : T
    T3 = !isbinary && T2 == Vector{UInt8} ? String : T2
    T4 = date_and_time && T3 <: DateTime ? DateAndTime : T3
    return notnullable ? T4 : Union{Missing, T4}
end

Regards, Pascal.