Thanks for the quick reply. That is what would be my next step. I don't like it because this component set seems to be working fine and it is likely something I've done that broke it.
On Sat, May 11, 2013 at 8:05 AM, Olivier Sannier <obo...@free.fr> wrote: > Hello, > > I've seen this happen with other component sets because the components as > well. > The situation triggering this is yet unclear to me, but it is a > combination of putting the component on a form and letting the application > delete it. > It appears, but I'm not quite sure, that finalization of the main form is > done AFTER all units are finalized while it used to be the other way round. > To circumvent this, I set the global variable to nil and test for it in > the appropriate location. > So basically, to be on the safe side, ICS should set SslCritSect to 0 when > finalizing and test for this value in UnloadSSL. > > On 11/05/2013 15:42, John Reid wrote: > >> We are using the ICS in an application that emails various entities (not >> spam). Recently, we ran into an issue where the application would not shut >> down properly. On some systems it simply persists and on others it reports >> that an error occurred and it needs to close. This is after it has >> completed all of its functions. I traced into it and found the >> Finalization >> section of OverbyteICSWSocket was executing before the >> TComponent.DestroyComponent. The Finalization section calls >> DeleteCriticalSection(**SslCritSect). TComponent.DestroyComponent calls >> the >> destructor of all its components. Those destructors ultimately call >> UnloadSsl and this method uses EnterCriticalSection(**SslCritSect) and >> LeaveCriticalSection(**SslCritSect), which causes an access violation and >> prevents the app from closing. >> This problem has only been seen recently; prior versions of this app with >> this version of the ICS, the exact same components with the exact same >> properties, and only slight modification to the code do not hang like >> this. >> First, is executing the Finalization before calling the component's >> destructor the *normal* order? Is there some way to control this order? >> Second, has anyone else seen this issue? Is there any way to *fix* this >> short of removing the use of the SslCritSect in UnloadSsl? >> Thanks in advance for any help. >> >> > > -- > To unsubscribe or change your settings for TWSocket mailing list > please goto > http://lists.elists.org/cgi-**bin/mailman/listinfo/twsocket<http://lists.elists.org/cgi-bin/mailman/listinfo/twsocket> > Visit our website at http://www.overbyte.be > -- *John Reid* *Software Development Manager* 6415 South 3000 East Suite 150 Salt Lake City, UT 84121 Phone: 801.305.3995 Fax: 801.521.6222 *www.piracle.com* -- To unsubscribe or change your settings for TWSocket mailing list please goto http://lists.elists.org/cgi-bin/mailman/listinfo/twsocket Visit our website at http://www.overbyte.be
