lispOS and persistent store
Tue, 29 Apr 1997 18:02:24 -0700
David Gadbois wrote:
> Date: Fri, 25 Apr 1997 19:33:11 -0700
> From: Luca Pisati <email@example.com>
> I'm just trying to understand what are we trying to do here.
> It is totally up in the air. The are lots of exciting ideas floating
> around. Until someone sits down and actually does something, we are
> mapping out the possibilities and arguing the benefits of various
> approaches. At some point, we'll need to articulate clear goals so
> that we can get good work done and so newbies can get up to speed
> 1. A Lisp-Machine (a full featured Lisp based OS, with
> classic OS structure: file-system, network ....),
> running directly on some hardware, or lying on top
> of some kernel.
> This is what I would like to see. A single, object-oriented address
> space where every capability is just a pointer dereference or a
> function call away.
Yes, but, how can you solve problems as efficient tags,
without a tagged bit architecture, for instance ?
I worked on XL's and I do work now on commercial computer
graphics software (90% in CL/CLOS/MOP) running under ACL on IRIX.
We do sell our software, and since it is for real-time market,
we have to run it fast and efficiently. I spent a lot of time
optimizing code to avoid "boxing" of single-floats into a tag +
a pointer + a single float ... and so on ...
Can you really build a pure LispOs without a tagged architecture,
and then be prone to Lisp Objects inefficiency ?
Can you write efficient code not only for A.I.sh kind of software,
but also for numerically-heavy applications ... ?
> 2. A Virtual Lisp Machine, running on top of existing
> OS and hardware.
> I'd like to view this as more of an engineering approach than a
> fundamental design issue. As existence proofs, we have Open Genera
> and the various Smalltalk environments: You hardly have to know that
> there is some kind of hypervisor between you and the metal.
Isn't it Open Genera emulating a 40bits (8 tag + 32 data) hardware
on top of a 64 bits machine (then wasting 24bits per word) ?
> 3. A Lisp based OS with radically new approaches as
> substituting persistent objects to file system and so on ...
> This is the sizzle that makes the whole project so appealing: The
> very ability to be able to consider such new ideas as "simple matters
> of programming" rather than as major political and organizational
The only problem is that compromises accepted for the sake of 1) and 2)
could make difficult (if not impossible) to maybe reach goal 3).
Luca Pisati Voice: (310) 577-0518
Nichimen Graphics Fax: (310) 577-0577
12555 W. Jefferson #285 EMail: firstname.lastname@example.org
Los Angeles, CA 90066 Web: http://www.nichimen.com