Date: Thu, 15 Jun 1995 08:46:15 -0400
From: kagel AT quasar DOT bloomberg DOT com
To: egdorf AT zaphod DOT lanl DOT gov
Cc: djgpp AT sun DOT soe DOT clarkson DOT edu
Subject: Re: Mapping files to memory?
Reply-To: kagel AT ts1 DOT bloomberg DOT com

   Xref: news-dnh.mv.net comp.os.msdos.djgpp:374
   From: egdorf AT zaphod DOT lanl DOT gov (Skip Egdorf)
   Newsgroups: comp.os.msdos.djgpp
   Date: 13 Jun 1995 16:48:02 GMT
   Organization: Los Alamos National Laboratory
   References: <DA45p3 DOT CMy AT jade DOT mv DOT net>

   In article <DA45p3 DOT CMy AT jade DOT mv DOT net> kagel AT quasar DOT bloomberg DOT com writes:

      For those not familiar with memory mapped I/O: I believe the facility first
      appeared on DEC10 and DEC20 systems under TOPS10/20.

   Ahhh, the nit-pick of the day. This was (as with so much in so many
   systems) inspired by Multics, where memory mapping was the
   norm. Segments (files on multics) were just parts of your virtual
   memory that could be made visible by mapping them. Lots of systems
   have lifted memory mapping because it is such a "right thing" for so
   many problems.  Multics just did it in a more complete way because it
   was a basic concept from the beginning.

					   Skip Egdorf
					   hwe AT lanl DOT gov

Thanks Skip.  I had a nagging suspicion when I wrote that response that Multics
was the original source of mapped I/O, but I was only certain of the DEC 10
implementation as preceeding UNIX and VMS implementations, hence the "I
believe" hedge there.

-- 
Art S. Kagel, kagel AT ts1 DOT bloomberg DOT com

Variety is the soul of pleasure.  --  Aphra Behn