> Interesting. Generally, I allocate cursors exactly at the same time as I open
> transactions;
> not sure if this works with the mysql connector, but with psycopg2
> (PostgreSQL), my code looks like this:
>
> with conn, conn.cursor() as cur:
> cur.execute(...)
> ... = cur.fetchall()
>
>
On Mon, Oct 24, 2016 at 3:46 AM, Joseph L. Casale
wrote:
> It really is that simple which is why I am baffled. Given the throughput is so
> low, if I close the cursor and connection at the end of loop and instantiate
> them
> both at the start of the loop, it works as expected but that's obviousl
> Perhaps you simplified too much, but changes between the select and the
> update could be lost. I think you need at least three states:
>
> 1 mark rows where baz is null (by setting baz to some value other than NULL
> or 42, 24, say: set baz = 24 where baz is NULL)
> 2 show marked rows (select
Joseph L. Casale wrote:
> I have some code that I am testing on Windows without c extensions which
> runs on a RHEL server with c extensions. In a simplified test case as
> follows:
>
> connection = mysql.connector.connect(...)
> cursor = connection.cursor(cursor_class=MySQLCursorDict)
> while Tr
