files, printers, etc.

Marc Wachowitz mw@ipx2.rz.uni-mannheim.de
Mon, 5 May 97 17:20:51 +0200


"Chris Bitmead uid(x22068)" <Chris.Bitmead@alcatel.com.au> wrote:
> There shouldn't just one set method for accessing an object (like the
> /dir/dir/dir/file of UNIX). In fact going down levels in the file
> system might be function calls rather than a simple traversal of
> layers.

I suggest looking at Inferno's name space handling for something which
is both sufficiently similar to a traditional file system to provide
a comfortable integration with existing software, and sufficiently
flexible to allow very different models to be implemented on it without
cutting off old applications. Nothing prevents us from having higher-
level services which support such an interface and wrapping lower-level
services in a higher-level framework, but I doubt the "all arbitrary
first-class Lisp objects" approach would go very far wrt. integration
of other software.

IMO, most information should be accessible and maintainable with a
text based interface, as long as there's no horrible conceptual gap
implicit in the information, and nobody except a system programmer
should need to know which representation (if any) is "primary". Even
if it's "Lisp down to the core", such abstractions are worth having.

(Unfortunately, I don't remember and currently can't look up the exact
location of the Inferno descriptions, but I think it's reachable from
http://www.lucent.com, with a title of "Technical Reports" or similar.)

-- Marc Wachowitz <mw@ipx2.rz.uni-mannheim.de>