[comp.windows.ms.programmer] UAE problem

peter@trlamct.trl.oz.au (Peter Campbell) (06/25/91)

Could someone please help me with the following problem:

Here at my company we have a program, written for Windows, that does some
fancy graphics viewing, shows some text, etc.  It is basically a
demonstration program.

If you don't touch it it will cycle through a predefined set of actions -
opening windows, showing graphics, closing windows, etc.  However, after
about thirty minutes Windows will kill the program, complaining with an:
"Unrecoverable Application Error","Terminating Current Application".
There is no suggestion to reboot the machine as normally happens, and
we can run the process again without rebooting.  Windows doesn't hang
the machine soon after - the only problem is that the program will still
terminate again after a period of time.

From what I've been reading in the discussion lately concerning problems
with Windows because it doesn't free up resource handles (or something
like this), it sounds like this could be the problem.  That is, the
program starts a process, the process is carried out, the program thinks
that it has given back the resources, and continues, until Windows says
it has run out of resources.  Even though there is plenty of memory etc.
left and the program thinks nothing is wrong, Windows complains, as it
doesn't think there are any of these weird resource handles left.

Does this sound right?  If so does anyone know if there is any way around
it, e.g. some command which will free up the handles properly, or trick
Windows into doing it?  Is this problem meant to be fixed in Windows 3.0A,
plus when is the new Windows meant to come out?  (I thought it was originally
scheduled for late last year, then early this year, then a few week ago,
then ???)

Another variable to add to the puzzle:  We had codeview from the SDK set
up on a separate monitor, so we could try stepping through the program
and try to work out what was going on.  However, the problem failed to
appear when this technique was used - it didn't matter if we single-stepped
or just set the program to run normally (albeit a bit slower) and left
it going overnight.  Hence it appears codeview is either catching the
problem before it can affect Windows and not telling us, or it is calling
the Windows functions in an abnormal manner which works better.  So there
appears that there is some way to get around the problem, as codeview
manages it.  Or perhaps it is some timing problem, which codeview 'fixes'
merely because it is altering the timing?

This is really a problem.  If it can be solved, either by some patch or the
new Windows, then we have a product which can be further developed and
maybe has commercial applications.  If not then we have to scrap the whole
idea and give up Windows in disgust.

Any help via either Email or in this newsgroup would be much appreciated.  I
will summarise any help I get via Email (if any) and send in the results to
here - I'm sure that there must be other people out there who have had the
same or a similar problem (or are we just lucky? ;-).

Thanks in advance.

Peter K. Campbell
Email:  p.campbell@trl.oz.au
#include Ye.Olde.Standard.Disclaimers.And.Notices.Of.Copyright.etc