I know that the data pump itself is not responsible for creating an
additional thread.

Are you using event alerters any?

This sounds to me like a perfect case where events and signals with a
suspended thread in its own connection context would be in order.

FWIW,
Jason Wharton
CPS - Mesa AZ
http://www.ibobjects.com

----- Original Message -----
From: <lester@...>
To: "IBO" <IBObjects@egroups.com>
Sent: Wednesday, January 24, 2001 10:35 AM
Subject: [IBO] Memory Leak


> I am getting out of my depth here so a pointer would help.
>
> In order that I can carry on processing messages as soon as they arrive
> ( one is the time signal for the system ) I have created a list of
> records that have to be updated, and these are processed in order. So I
> have a routine that uses IB_DataPump to copy a record from the main
> server to the local copy. It is running on a simple timer, and uses the
> same timer to check for a stuck update. If the list is empty, then I
> either just stop scanning and wait for the next number on the list, or I
> drop the connection, and remake it later.
>
> The logic works fine, and everything 'feels' better responce wise, but I
> have a growth problem that I took some time to track. I have Memproof
> running now, and while it has a problem running on W2k, crashing when
> trying to load gds32.dll, NT and 98 are fine if I have another
> application already using gds32. Memproof is logging a 'Create Thread'
> for every update, hence the growth. I have a separate module with it's
> own connections and timer which should be running all the time, so where
> is the extra thread coming from, or more to the point, how do I track
> this. Initial attempts are pointing at IB_DataPump, but there are two of
> them, and only one thread.
>
> Can anybody help?
>
> --
> Lester Caine
> -----------------------------
> L.S.Caine Electronic Services
>
>
>