From: Bill Stewart <stewarts@ix.netcom.com>
To: cypherpunks@toad.com
Message Hash: b9a804da6b807b4df7959b9868c772a8546b9e5f697ce962367ad8c05f44e1b9
Message ID: <3.0.3.32.19971116125402.007134ec@popd.ix.netcom.com>
Reply To: <199711160027.BAA16314@basement.replay.com>
UTC Datetime: 1997-11-17 00:17:00 UTC
Raw Date: Mon, 17 Nov 1997 08:17:00 +0800
From: Bill Stewart <stewarts@ix.netcom.com>
Date: Mon, 17 Nov 1997 08:17:00 +0800
To: cypherpunks@toad.com
Subject: RAM disks for temp files?
In-Reply-To: <199711160027.BAA16314@basement.replay.com>
Message-ID: <3.0.3.32.19971116125402.007134ec@popd.ix.netcom.com>
MIME-Version: 1.0
Content-Type: text/plain
At 01:27 AM 11/16/1997 +0100, Anonymous wrote:
>> (Hint, temporary files all over the place.)
> For you old farts who have not been out in the real world for a
>while, you should make note of the fact that the price of memory
>has dropped, and it is now feasible to implement RAM disks to
>store temporary files.
I was using RAM disks on my 386 machines with Win3.1.
Now that I've got faster machines with more RAM,
I'm running Win95 and can't find a ramdisk program -
does anybody know where to find one?
On Unix, you _could_ do things with temp files,
but it usually made much more sense to structure your programs
to use pipes, which means there's no temp file and only
a few KB of RAM buffering in between each program.
PGP's DOS heritage doesn't encourage this sort of programming,
though the new design for one-pass processing may have
made it either possible or unnecessary.
On machines with real operating systems, designing a
ramdisk includes deciding whether to make it virtual memory
that might get swapped if necessary, or nail it into RAM.
Since RAMdisks are usually intended for increasing speed,
it usually makes more sense to let the vm manager decide
what pages to page out and what pages not to,
but obviously if you're using it for security that's different.
SunOS had the /tmpfs file system design which let you
get hybrid behavior - temp files stay in RAM until
they need to get swapped out to disk, and most programs
delete them before that ever happens - really speeds up compiles.
>From a cryptographic standpoint, it's not ideal, since occasionally
your files would end up on disk, but they'd usually be safe.
Thanks!
Bill
Bill Stewart, stewarts@ix.netcom.com
Regular Key PGP Fingerprint D454 E202 CBC8 40BF 3C85 B884 0ABE 4639
Return to November 1997
Return to ““William H. Geiger III” <whgiii@invweb.net>”