Should be no handle left. But pay attention that if you destroy any
control
with a window handle from a thread which has not created the control,
then
the window handle is /not/destroyed (DestroyWindow fails but usually his
return value is not checked).
Everything is created in the
- Original Message -
From: Francois PIETTE [EMAIL PROTECTED]
Everything is created in the threads execute, so that should be fine.
You have to destroy also at the end of the execute method or from a
function
called from there.
Sure, but that is not the problem.
When a desktop
Everything is created in the threads execute, so that should be fine.
You have to destroy also at the end of the execute method or from a
function
called from there.
Sure, but that is not the problem.
When a desktop change is detected, these steps are executed:
- destroy HttpCli
-
Francois,
Paul, you told me in your intial message that having a handle still
allocated prevent desktop switching. As I know you are using threads, I
just
wanted to remind you that to destroy a window handle, you must call
Not quite:
I wrote Are there any handles left after freeing the
Hello,
IQRP 1.3.5beta has a serious problem: when a client in Russia who has up to
200 concurrent users all the time leaves the server for 1-2 hours, it hangs
with 0% CPU usage in such a way that he cannot stop the service OR kill the
process from task manager (it says Access denied). I wonder
Fastream Technologies wrote:
Hello,
IQRP 1.3.5beta has a serious problem: when a client in Russia who has
up to 200 concurrent users all the time leaves the server for 1-2
hours, it hangs with 0% CPU usage in such a way that he cannot stop
the service
Sounds like some wait routine waits
Thanks for the reply. This is a service application, not a service. I
wonder why Eurekalog cannot produce any error report for this--it has
an option for freeze.
On 1/27/07, Arno Garrels [EMAIL PROTECTED] wrote:
Fastream Technologies wrote:
Hello,
IQRP 1.3.5beta has a serious problem: when
Fastream Technologies wrote:
Thanks for the reply. This is a service application, not a service.
What do you think is the difference between a service application and a
service? Hasn't a service application at least one service anyway?
I
wonder why Eurekalog cannot produce any error report
- Original Message -
From: Arno Garrels [EMAIL PROTECTED]
To: ICS support mailing twsocket@elists.org
Sent: Saturday, January 27, 2007 9:12 PM
Subject: Re: [twsocket] Crash (hang) under heavy load after 1
hourproblemforserver service
Fastream Technologies wrote:
Thanks for the reply.
Hello Fastream,
Something specific to client's server I guess. Is there a way to see which
line number did it got stuck without installing the compiler there?
I suggest a lot of debugging code in a file. Put it between a global
#define do that you can remove it with just one line in comment.
Fastream Technologies wrote:
The problem is when I try here on my Win XP/Vista's it does not occur.
Something specific to client's server I guess.
Using threads can be very exiting ;-)
Is there a way to see
which line number did it got stuck without installing the compiler
there?
Arno,
You are wrong in saying that service applications are not killable from
task manager. They are when everything is ok/under normal circumstances. Yet
a special kind of error prevents this behavior that I am unable to reproduce
here. They say they can only stop the dead service by
12 matches
Mail list logo