Fetchall = False reopens connection

Guest · Post by **Guest** » Wed 27 Apr 2005 09:38

> Is this the right solution?

Possible. Try to switch to MyConnection.Options.Direct := False mode and repeat a test.

lsc82 · Post by **lsc82** » Thu 28 Apr 2005 15:16

The Direct option is not the problem.

I've found the bug. It is not a real bug of MyDAC, it only reaches the limit of 5000 in Windows XP, 2003.
It is a known bug of Windows XP, see Microsft KB:

http://support.microsoft.com/kb/q196271.htm

How to reproduce:

- On a Windows XP machine
- a program that simply open and close 10000 a TMyQuery with FetchAll = False, without any other operation:

Code: Select all

   for j := 1 to 10000 do
      begin
      q.Open;
      q.Close;
      end;

- the "Open" statement will raise a connection error after about 3000-4000 cycles.

I've applied the solution proposed in Microsoft's articol, and the problem disappear.

At now, I'm proposing you a way to optimize your use of fetchall = False.
I think you miss a method to "change" the sql statement of a query without closing the "swap" connection (this superfluous operating raise the problem).
I've try to modify you source introducing the method

procedure TMyQuery.ChangeSQL(SQLStatement: String);
begin
FDontCloseSwap := True;
SQL.Text := SQLStatement;
Open;
FDontCloseSwap := False;
end;

and modify your ReleaseSwapConnection method

procedure ReleaseSwapConnection;
begin
if FDontCloseSwap then Exit;
[...]
end;

and modify your QuerySwapConnection method

procedure QuerySwapConnection;
begin
if FDontCloseSwap and Assigned(FConnectionSwap) then Exit;
[...]
end;

So your component does not reopen the swap connection at every query change without a real need to do this.

What do you think about this?

Thanks you for your component. Is really great and fast!

Ikar · Post by **Ikar** » Thu 28 Apr 2005 15:25

>> Is this the right solution?

I quite doubt, it must raise "Command out of sync" error.

By the way, we couldn't reproduce the problem with 4.1.10 + WinXP.
Please send us a complete small sample.

Ikar · Post by **Ikar** » Fri 29 Apr 2005 06:37

> I've found the bug.

We are not sure that the reason of the problem is in it.

> How to reproduce:

This code doesn't cause an error at us.

> So your component does not reopen the swap connection at every query change
> without a real need to do this.

We are planning to optimize this part but taking on account FetchAll=False restrictions not at the nearest time.