Date: Wed, 2 Nov 94 15:49:53 EST
From: dliu AT faraday-gw DOT njit DOT edu (Dong Liu)
To: DJGPP users list <djgpp AT sun DOT soe DOT clarkson DOT edu>
Subject: Re: C++ misc suggestions / queries
References: <9411011021 DOT AA18045 AT hpas3 DOT prl DOT philips DOT nl>
	<Pine DOT SOL DOT 3 DOT 90 DOT 941102084907 DOT 29035A-100000 AT chopin DOT udel DOT edu>
Mailer: VM 5.32 (beta) for GNU Emacs 18.59.5

>>>>> On Wed, 2 Nov 1994 08:53:23 -0500 (EST), Ed Phillips <flaregun AT udel DOT edu> said:
> > 
> > >	DEVICE=C:\DOS\EMM386.EXE NOEMS X=C800-C9FF
> > >
> > > Commented this out and demacs came up fine!  Now... what have I lost
> > > by doing this?
> > 
> > Why would anybody working with DOS 6.x put ``NOEMS'' on the EMM386
> > line?  I would say it's a left-over from DOS 5.0, where Expanded
> > and Extended memories (sic) were managed with two distinct pools,
> > and so whoever wanted max XMS had to be deprived of EMS.  In DOS
> > 6 this is no more the case, AFAIK.  Do I miss something here?
> > 
> 
> I would replace 'NOEMS' by 'RAM FRAME=NONE'.
> You end up with the same amount of upper memory (blocks)
> and still provide EMS for programs that need it.


Ed>   I have wondered about this FRAME=NONE.  What do you lose by doing 
Ed> this?  Can EMM386 use XMS to store the page frame in this case?  I use 
Ed> Lightning/CD disk caching software (which works great with DJGPP) and it 
Ed> could use some more upper memory.  With EMM386 it ends up having to put 
Ed> part of itself in low mem (40k).

From my own experience, Codeview (I forgot the ver, the one comes with
VC++) refuses to work if FRAME=NONE is used.

--dong