Early file system layouts (was: Splitting / and /usr)

Early file system layouts (was: Splitting / and /usr)

[norman]

Warren is right that even the First Edition manual says that init
mounts /usr, implying that /usr was a distinct file system even that
early.  It seems to me that the original question Greg forwarded
from the NetBSD list was also after when /usr/bin appeared, which
isn't necessarily the same date.

A possible answer from old manuals:

- Second Edition sh(I) (dated 3/15/72):
	If the first argument is the name of an executable file,
	it is invoked; otherwise the string "/bin/" is prepended
	to the argument.  (In this way the standard commands,
	which reside in "/bin", are found.)  If the "/bin" file
	exists, but is not executable, it is used by the shell
	as a command file.

- Third Edition sh(I) (dated 1/15/73):
	If the first argument is the name of an executable file,
	it is invoked; otherwise the string "/bin/" is prepended
	to the argument.  (In this way most standard commands,
	which reside in "/bin", are found.)  If no such command
	is found, the string "/usr" is further prepended (to give
	"/usr/bin/command") and another attempt is made to execute
	the resulting file.  (Certain "overflow" commands live in
	"/usr/bin".)  If the "/usr/bin" file exists, but is not
	executable, it is used by the shell as a command file.

Notice the odd detail that non-executable files in /bin (early on)
or /usr/bin (later) get special treatment.  Does this mean that
shell scripts that weren't in /usr/bin had to be invoked explicitly
via `sh script' instead of just `script'?

Even deeper historic trivia: it occurred to me to check the fragments
of the PDP-7 system I have on paper to see whether /usr existed then.
I was quickly reminded that it almost certainly didn't because subdirectories
weren't really used then; there were no pathnames in that system.
(You could open only files in the working directory, though you could
link from another directory.)  When asked to invoke `x', the shell first
tried to open `x', then to link `x' from directory `system' and open the
result.  (Presumably it remembered to remove the needless link after the
open, but I'm not quite certain; the old paper copy is missing a few
lines just there.)  So even the name `bin' doesn't date back quite to
the beginning.

Norman Wilson

Received: (from major at localhost)
	by minnie.cs.adfa.edu.au (8.9.3/8.9.3) id KAA39649
	for pups-liszt; Tue, 18 Apr 2000 10:10:53 +1000 (EST)
Received: from henry.cs.adfa.edu.au (henry.cs.adfa.edu.au [131.236.21.158])
	by minnie.cs.adfa.edu.au (8.9.3/8.9.3) with ESMTP id KAA39645
	for <pups at minnie.cs.adfa.edu.au>; Tue, 18 Apr 2000 10:10:42 +1000 (EST)
Received: (from wkt at localhost)
	by henry.cs.adfa.edu.au (8.9.2/8.9.3) id KAA42226
	for pups at minnie.cs.adfa.edu.au; Tue, 18 Apr 2000 10:10:41 +1000 (EST)
	(envelope-from wkt)
From: Warren Toomey <wkt@cs.adfa.edu.au>
Message-Id: <200004180010.KAA42226 at henry.cs.adfa.edu.au>
Subject: Re: Early file system layouts
In-Reply-To: <200004172217.IAA38973 at minnie.cs.adfa.edu.au> from "norman at nose.cs.utoronto.ca" at "Apr 16, 2000  2:43:14 pm"
To: 
Date: Tue, 18 Apr 2000 10:10:41 +1000 (EST)
Cc: pups at minnie.cs.adfa.edu.au
Reply-To: wkt at cs.adfa.edu.au
X-Mailer: ELM [version 2.4ME+ PL43 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-pups at minnie.cs.adfa.edu.au
Precedence: bulk

In article by norman at nose.cs.utoronto.ca:
> - Second Edition sh(I) (dated 3/15/72):
> 	If the first argument is the name of an executable file,
> 	it is invoked; otherwise the string "/bin/" is prepended
> 	to the argument.  (In this way the standard commands,
> 	which reside in "/bin", are found.)  If the "/bin" file
> 	exists, but is not executable, it is used by the shell
> 	as a command file.
> Notice the odd detail that non-executable files in /bin (early on)
> or /usr/bin (later) get special treatment.  Does this mean that
> shell scripts that weren't in /usr/bin had to be invoked explicitly
> via `sh script' instead of just `script'?

Can't tell, we don't have the source code. In the Nsys kernel (dated
just before the 4th Edition), files must have the execute bit on or
they can't be exec(2)d.

	Warren