Can I use multiple cursors on one connection with pyodbc and MS SQL Server?

Question

I'm using pyodbc on python 2.6 to connect to Microsoft SQL Server 2005. I open a connection, create a couple of cursors:

c1 = connection.cursor()
c2 = connection.cursor()

and then run a query on the first cursor.

c1.execute("select * from foo")

Now I run a query on the second cursor:

c2.execute("select * from bar")

...and I get an error: "Connection is busy with results for another hstmt."

After I do a c1.fetchall() or c1.close() then I can use c2.

My question is: Why am I even allowed to create multiple cursors on a connection, if I'm only allowed to use one at a time, and the same one can always be reused? And, if I want to run a query for each row of the results of another query, like this:

for x in c1.execute(...):
    for y in c2.execute(...):

do I really have to create multiple connections to the same database?

Answer 1

This appears to be supported via multithreading: http://technet.microsoft.com/en-US/library/ms131700(v=sql.90).aspx

Answer 2

According to this guy

Cursor objects are used to execute SQL statements. ODBC and pyodbc allow multiple cursors per connection, but not all databases support this.

and you can determine concurrent cursors can be supported with:

import pyodbc
connection = pyodbc.connect(...)
how_many = connection.getinfo(pyodbc.SQL_MAX_CONCURRENT_ACTIVITIES)
print(how_many)

Answer 3

I my own practice I have never met necessity to use more than one database cursor. Such problems are used to be solved by sophisticated SQL queries (joins, groups). Or (if you can to ignore perfomance issues) by using several simple queries.