Julia ReadOnlyMemoryError with ODBC SQL query - sql

Question

I'm working on writing a function that queries a SQL database and I'm encountering a ReadOnlyMemoryError() that has me stumped. The issue is that when I run my code as a simple script everything functions as expected. But when I try to wrap the exact same code in a function I get the ReadOnlyMemoryError().
Here's the script version of my code:
using ODBC
using DBInterface
using Dates
using DataFrames
server = "server string "
username = "username "
password = " password"
db = " db name"
start_date=Nothing
end_date=Nothing
if start_date == Nothing || typeof(start_date) != "Date"
start_date = Dates.today() - Dates.Day(30)
end
if end_date == Nothing || typeof(end_date) != "Date"
end_date = Dates.today()
end
query = """ SQL SELECT statement """
connect_string = "DRIVER={ODBC Driver 17 for SQL Server};SERVER=" * server *
";DATABASE=" * db *
";UID=" * username *
";PWD=" * password
conn = ODBC.Connection(connect_string)
df = DBInterface.execute(conn, query) |> DataFrame
This works as expected, the result is a dataframe df with about 500k rows. However, when I try use this same code to make a reusable function I get the error:
using ODBC
using DBInterface
using Dates
using DataFrames
function get_cf_data(start_date=Nothing, end_date=Nothing)
server = " server string "
username = " user name"
password = " password"
db = " db "
if start_date == Nothing || typeof(start_date) != "Date"
start_date = Dates.today() - Dates.Day(30)
end
if end_date == Nothing || typeof(end_date) != "Date"
end_date = Dates.today()
end
query = """ SQL SELECT statement """
connect_string = "DRIVER={ODBC Driver 17 for SQL Server};SERVER=" * server *
";DATABASE=" * db *
";UID=" * username *
";PWD=" * password
conn = ODBC.Connection(connect_string)
df = DBInterface.execute(conn, query) |> DataFrame
return df
end
In this case, when I try to call from REPL get_cf_data() I get ERROR: ReadOnlyMemoryError(). I'm somewhat new to Julia, so any insight would be very much appreciated. Thank you!

Answer 1

As commented, best practice in most programming languages when integrating APIs like ODBC connections is to close and release its resource after usage.
Additionally, consider parameterization (best practice in any language running SQL that passes literal values) where you set up a prepared SQL statement and bind values in a subsequent execute call.
function get_cf_data(start_date=Nothing, end_date=Nothing)
server = " server string "
username = " user name"
password = " password"
db = " db "
if isnothing(start_date) || typeof(start_date) != "Date"
start_date = Dates.today() - Dates.Day(30)
end
if isnothing(end_date) || typeof(end_date) != "Date"
end_date = Dates.today()
end
# PREPARED STATEMENT WITH QMARK PLACEHOLDERS
sql = """SELECT Col1, Col2, Col3, ...
FROM myTable
WHERE myDate BETWEEN ? AND ?
"""
connect_string = "DRIVER={ODBC Driver 17 for SQL Server};SERVER=" * server *
";DATABASE=" * db *
";UID=" * username *
";PWD=" * password
conn = ODBC.Connection(connect_string)
# PREPARE STATEMENT AND BIND PARAMS
stmt = DBInterface.prepare(conn, sql)
df = DBInterface.execute(stmt, (start_date, end_date)) |> DataFrame
DBInterface.close(stmt) # CLOSE STATEMENT
DBInterface.close(conn) # CLOSE CONNECTION
stmt = Nothing; conn = Nothing # UNINTIALIZE OBJECTS
return df
end