Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site rlgvax.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!bellcore!decvax!genrad!panda!talcott!harvard!seismo!rlgvax!guy
From: guy@rlgvax.UUCP (Guy Harris)
Newsgroups: net.arch
Subject: Re: Uses of rms
Message-ID: <514@rlgvax.UUCP>
Date: Mon, 25-Feb-85 02:28:00 EST
Article-I.D.: rlgvax.514
Posted: Mon Feb 25 02:28:00 1985
Date-Received: Thu, 28-Feb-85 01:17:01 EST
References: <109@rti-sel.UUCP>
Organization: CCI Office Systems Group, Reston, VA
Lines: 21

> When in a previous article I suggested that you bypass the operating system
> file system, I meant it. If you really need efficiency, then why don't you
> map your files to virtual memory. If the files are contiguous, then
> your "file operations" will be as fast as memory paging.

That doesn't bypass what really *should* be called the file system, namely
the code that turns references to block N of thus-and-such a file into
references to physical disk block M.  Some of the confusion here is being
caused by the fact that RMS and F11*ACP are being collectively referred to
as "the file system".  Well, 'taint so.  I'd call F11*ACP the file system.
It supports the creation of files, mapping of virtual blocks within a
file to physical blocks on a disk, and extension of files.  It maintains
the size of the file and various access/modification times.

Support of file structure is a separate function on RSX, VMS, and UNIX.
UNIX just makes this separation clearer than RSX/VMS does (or, at least,
UNIX documentation doesn't hide the raw interface to the file system as
well as RSX/VMS documentation does).

	Guy Harris
	{seismo,ihnp4,allegra}!rlgvax!guy