On Tuesday, 97/03/11, James Mckenzie wrote to David Noon about "Dpmi
Limits" as follows:
JM> Sorry to say, but it will crash OS/2.  And I know what you 
JM> are saying about the pointer.  Weird does not begin to 
JM> describe it.  Actually, it can address 32MB -1 byte.
Hi James,
Since you roused my curiosity, I decided to write a little test
program that would access 40MB of DPMI storage. It worked perfectly,
allocating and _committing_ all 40MB without incident beyond bloating
the swap file a bit.
JM>  DN> Addressable file? What's that?
JM> 
JM> The file that it can access.
I still don't understand this. DPMI is memory, not file.
JM>  DN> The limit should be determined by the DPMI server, not the client. 
n
JM>  DN> a VDM the server should be OS/2's VMM, not DOS/4GW.
JM> 
JM> You are correct according to the way things SHOULD work, 
JM> but remember, DOS/4GW was written to run under pure DOS, not an
JM> OS/2 VDM.
A DPMI client is obliged to obey the DPMI conventions. These are not
dependent on pure DOS. Indeed, using DPMI makes DOS very impure, since
a DPMI server switches the CPU from real mode to protected mode.
The DPMI server should provide the standard interface too. The VMM in
an OS/2 VDM does precisely this. So does DOS/4GW server, when it is
active. It isn't active in a VDM. Only the client code from DOS/4GW is
active in a VDM.
Regards
Dave

 * KWQ/2 1.2i * "... and Morgo, as the friendly drelb." -- G. Owen
--- Maximus/2 3.01
---------------