>First, in design, if you fail to release the mutex (via the functions I mentioned previously), you will have to close and re-start VFP in order to do further testing. This is because VFP is the thread that created the mutex, not the VFP application. Simply exiting an VFP application in design mode will not destroy the mutex.
>

The example was mean to test that the system does release the mutex, and I said that you should build a executable.

>Further, failing to do is, IMHO, bad design. Any application should, take steps restore the state of the environment to that which it originally found it. Relying on Windows to clean up after something that I (the programmer) did, is simply a bad design choice.

Agreed. But, if for some reason this cleaning code doesn't get called because some unexpected error, you still can start the application again.

>
>Second, if a thread terminates abnormally via a GPF, the mutex will not be destroyed. This is not subject to debate because I've experienced it first hand. Be advised that code was included in the application shutdown to release and destroy the mutex, and it remained.

I need to be able to intentionally generate a GPF to test this.
OTOH, If someone get a GPF in Windows 9x, I think it would be recommended to reboot.