Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!samsung!usc!csun!kithrup!sef
From: sef@kithrup.COM (Sean Eric Fagan)
Newsgroups: comp.unix.internals
Subject: Re: Wizard-level questions
Message-ID: <1991Jan27.092637.3040@kithrup.COM>
Date: 27 Jan 91 09:26:37 GMT
References: <16048@sdcc6.ucsd.edu> <1991Jan26.142403.22812@mp.cs.niu.edu>
Organization: Kithrup Enterprises, Ltd.
Lines: 26

In article <1991Jan26.142403.22812@mp.cs.niu.edu> rickert@mp.cs.niu.edu (Neil Rickert) writes:
> I hope not.  Otherwise permissions on directories wouldn't do much.  I
>do think the system design would have been cleaner if you only accessed
>by i-node number, and mapping filename to inode was done outside the kernel.
>But I doubt that I have many supporters in this "keep the kernel small" view.

On the contrary, you actually have quite a few supporters of that view.  But
your approach is wrong, for the reason you pointed out.  (Although... I
believe that at least one version of SunOS allowed any user to access a file
they had permission to, no matter what the permissions of the intervening
directories were.  That is, /home/sef/.private/foo, assuming foo readable by
all, and .private mode 700, would still have allowed foo to be read by
anyone who knew of it.  And this is a *bug*!  [I really hope it's been
fixed, or, better yet, that I imagined the whole thing.])  Doug Gwyn, I
think it was, had a system for "file handles," which are somewhat nice, but
are mostly an extension of the currect file-descriptor theme.

Note that none of the current setup requires that the filesystem go into the
kernel.  You can have the filesystem be a seperate process, if you wish.
(And some of us wish... 8-))

-- 
Sean Eric Fagan  | "I made the universe, but please don't blame me for it;
sef@kithrup.COM  |  I had a bellyache at the time."
-----------------+           -- The Turtle (Stephen King, _It_)
Any opinions expressed are my own, and generally unpopular with others.