>Latest update...
>I just received customer config.sys and autoexec.bat.
>Numerous device drivers loading. Emm386 and himem device lines. CDRom device drivers loading. Anti-virus in autoexec.bat loading.
>I don't know whether safe mode by-passes those in one of our tests. I had him move both files to another directory and we are going to test again.

Safe Mode bypasses both AUTOEXEC.BAT and CONFIG.SYS; the odds are good that you'll be able to run without most of these with a fresh install of Win95 or Win98, unless he specifically needs the real-mode CD-ROM driver.

I'd try to run with just a FILES= statement in the CONFIG.SYS, and let Win95 handle the memory management itself without loading up HIMEM and EMM386 explicitly (Win95 will load HIMEM, and will use UMB space for itself without EMM386 helping things along), unless he has apps that absolutely require EMS.

>
>Notes:
>1) he can run the software once for as long as he wants with no problem. Exit the program and reenter he will always freeze up his machine.

This is a pure SWAG, but try this in place of a QUIT statement:

DECLARE ExitProcess IN WIN32API INTEGER uExitCode
=ExitProcess(0)

I had problems with some .DLLs not releasing properly under VFP 3 and 5 (I seem to do a lot of DECLARE...DLL stuff in my code, and it generally isn't safe for me to issue a CLEAR DLLS); I started using ExitProcess() in place of QUIT, and the problems cleared up. I never tracked down exactly which DLL declarations were causing me problems under VFP 5 since this worked well, and it still works under VFP 6. It sounds like you're starting something, maybe an OLE Server, that is not shutting down, or not shutting down properly, when you quit the first time, and you're dying when you go to start it a second time.

What, if any, ActiveX components are you using in your apps? Not just controls; do you start any out-of-process servers for your app, and is there any way to tell if they're still running after your app quits? I've seen this scenario with some poorly-conceived out-of-process servers (I should know, since I'm the one who wrote them) that died in a particularly ugly and unforgiving fashion if a second instance started on the same system.

>2) We ran windows resource meter before and after first time in software. No resource losses.
>I'll let you know about the test after I hear from him...
>Thanks for the continuing support.
>
>Robert