PS> I create a 'access-handleing-thread', that as above receives messages via 
 PS> queues or pipes (probably queues, since pipes' point-to-
 PS> point), and then send 
 PS> messages to other threads that requested the access. 
 PS> This thread will then 
 PS> have the access until it sends
"i'm-finished-let-the-next-thread-get-the-
 PS> access" to the access-handleing-thread, that then if 
 PS> there is one, passes the 
 PS> access on. But this too can get complicated.

I'm not very well versed on this sort of thing (I still need to find a OS/2
C++ compiler first! =) but I see one problem with that.  The imfamous
Windows 3.1 bug, it sounds like you're suggesting non-preemptive
multitasking.  What if the thread that currently has access dies?  Will
everything then be locked out of that file?  Also, what if the thing that
wants access, say a tosser, works for 5 minutes at it.  That could feeze
everything else until it's done.  I'm not sure how Maximus does it, but in
the msgapi, there's a bit about how it locks things.  And somehow it does
to pre-emptive multitasking, because while squish is tossing you can access
the message areas.  Again (as I said in the C++ echo), the commenting may
be horrible, but the source for these programs may be worth looking at to
see how Scott solved these problems.

I know this doesn't help much, but it does point out two weaknesses...I
only wish I had some better ideas for you...good luck!

--- Maximus/2 3.00