From: Charles Sandmann <sandmann AT new-orleans DOT NeoSoft DOT com>
Subject: Re: Graphics and DPMI under Windoze
To: kunst AT prl DOT philips DOT nl
Date: Tue, 14 Feb 1995 08:34:56 -0600 (CST)
Cc: djgpp AT sun DOT soe DOT clarkson DOT edu

> I was wondering how you (Charles, DJ etc) have done/are planning the support
> of graphics in V2.0. Maybe you could post some remarks to the list.
> An obvious question is: will unpaged video memory access be supported ?
> (or only by the way as presented in my previous posting)

Csaba has a working version of GRX 2.0 which works under DPMI in alpha 
test right now.  Since it works with either V1.x or V2.x, it may be 
released before the entire V2 package.

As far as unpaged video memory access, I am somewhat at a loss to know
what to do.  I have an ET4000 card on an ISA bus, but have 16Mb of memory
so I can't enable the linear mode to even test it.  If your card and
bus can agree on a linear address which is OK, then a linear buffer could
be established and accessed via far pointers.  There are also tricks that
can be used on *SOME* DPMI providers to access this with a simple 
pointer, but it won't work everywhere.  So, this is the problem.  Some
machines in some environments can get linear graphics, others can't.  If you 
program expecting this behavior your code won't work everywhere.  Since
we control CWSDPMI you could always "fix" this in DOS, but you don't have 
the option to do this under Windows, Win95, or OS/2.

In V2 do we want linear graphics bad enough to limit where the code to 
run, or do you bite the bullet and deal with the page swapping and
write code that will run under *ALL* environments?

The answer to that probably depends if you are writing code for yourself
only or plan to share it on lots of machines.  So, at this point I would
say that linear graphics is an unresolved issue.  Does someone want to
go to the trouble of writing this knowing it isn't univerally useful?