Dear Gentle Persons,
Does any know a LISP system that works on embedded processors? Is there a
cheap or free ware version of Embedded LISP. The only version I am aware of
runs on commercial research and special application robot vendor, but it runs
on systems with PC Class Processors or above.
Robert Posey
If you consider unmanned space ship control an embedded application,
check out the news at http://www.xanalys.com :-)
Robert
Robert Posey wrote:
>
> Dear Gentle Persons,
>
> Does any know a LISP system that works on embedded processors? Is there a
> cheap or free ware version of Embedded LISP. The only version I am aware of
> runs on commercial research and special application robot vendor, but it runs
> on systems with PC Class Processors or above.
>
> Robert Posey
In article <·················@raytheon.com>, ·····@raytheon.com says...
> Does any know a LISP system that works on embedded processors?
Which processors? Which OS?
Ben Goetter wrote:
>
> In article <·················@raytheon.com>, ·····@raytheon.com says...
> > Does any know a LISP system that works on embedded processors?
>
> Which processors? Which OS?
It is really more of a matter of size and cost, first choice
a PIC, or HC11, or ARM,
second choice would be PowerPC Embedded Chip, which most likely does exist
at a high level at least.
Its more of a matter of not needing more than 16k -words and been able to
work over a serial interface from a PC.
Muddy
Robert Posey wrote:
>
> Ben Goetter wrote:
> >
> > In article <·················@raytheon.com>, ·····@raytheon.com says...
> > > Does any know a LISP system that works on embedded processors?
> >
> > Which processors? Which OS?
>
> It is really more of a matter of size and cost, first choice
>
> a PIC, or HC11, or ARM,
Not exactly Lisp, but a close derivative is NewtonScript, which is the
language of the Newton Messagepads. The StrongARM processor has an
amazingly brief code representation, and it is remarkably powerful for
its consumption.
NewtonScript is also heavily optimized for code brevity and portability.
For example it has a prototype-based object system, which means that new
objects can be created by referencing an existing one and storing only
slots that have different values from its parent. It allows
RAM-efficiency as well as flexibility, for example, you may have most of
your app in ROM, while retaining the flexibility to patch and add
functionality on the fly. Soups (a little in-memory OODB) are also
known to be space-efficient. There are several development tools for
NewtonScript (some of them work on the Newton itself).
As for PowerPC, Franz has a version of ACL for Linux and AIX for
PowerPC processors. ACL is already quite compact, and maybe some of its
features help (perhaps pure lisp libraries can be stored in ROM).
I don't know what processor Harlequin LispWorks was using on their space
flight.
Happy browsing!
Regards
Robert
In article <·················@raytheon.com>, Robert Posey
<·····@raytheon.com> wrote:
> Dear Gentle Persons,
>
> Does any know a LISP system that works on embedded processors? Is there a
> cheap or free ware version of Embedded LISP. The only version I am aware of
> runs on commercial research and special application robot vendor, but it runs
> on systems with PC Class Processors or above.
Have you looked for example at "L" from ISR?
Rainer Joswig, ISION Internet AG, Harburger Schlossstrasse 1,
21079 Hamburg, Germany, Tel: +49 40 77175 226
Email: ·············@ision.de , WWW: http://www.ision.de/
Robert Posey <·····@raytheon.com> writes:
> Does any know a LISP system that works on embedded processors?
You're probably referring to a different (less powerful) class of
processor and small memory, but since you said "embedded", I'll throw
in some information that is probably only of historical interest to you.
When Symbolics was in the business of integrated hardware/software
solutions, their ultimate hardware platforms were based upon a
proprietary chip family called "Ivory". This chip was used not only
in their workstation systems, but also in several "embedded" systems.
There were Ivory-based boards that plugged into Sun UNIX workstations
and into Macintosh personal computers, providing the full Symbolics
Genera ("Lisp Machine") development and delivery environment to the
host user and host applications. These "embedded" Genera boards
received I/O support (low-level network and disk interfaces) from
the host operating system, and employed standard connectivity
(TCP/IP services, X Windows for the user interface, and RPC)
for higher-level integration. You could think of these systems
as either a workstation-on-a-board, or as a symbol co-processor
(sort of like a floating point or array co-processor).
A later Symbolics product based on the Ivory chip was the Minima system.
This was a "pizza box" sized system designed to be embedded into real-time
applications; it was developed for AT&T for next-generation telephone
switching systems. Minima did not include the user interface or the
development subsystems of Genera (but of course it had all the benefits
and features of the Lisp system and the Ivory architecture).
Minima guaranteed (ie. GC) performance to meet the real-time requirements.
Minima was a different system from Genera, and it included and was
written in a Common Lisp (with language extensions needed for accessing
the hardware). Application development was using a cross-compiler on a
workstation running Genera, with remote debugging capability over TCP/IP.
Minima actually referred to the software part; I have forgotten the model
designation for of its particular pizza box. I can't even remember if
it included a disk and paging, or just a boat-load of memory.
It had multiple Ethernet ports to facilitate redundant network-based
applications. (Someone will chime in....)
Those systems were too big to fit in your microwave oven,
but Minima would have been suitable for some mobile robots.
>Robert Posey <·····@raytheon.com> writes:
>> Does any know a LISP system that works on embedded processors?
>
A simple application in scheme is described in :
[30] Jonathan Rees and Bruce Donald, "Program Mobile Robots in
Scheme", Proceedings of the 1992 IEEE International Conference
on Robotics and Automation, pages 2681-2688, 1992.
(mobot-scheme)
availiable at:
http://www.cs.cmu.edu/afs/cs/project/ai-repository/ai/lang
/scheme/txt/mobot.tgz
marc
--
------------------------------------------------------------------------------
email: marc dot hoffmann at users dot whh dot wau dot nl
------------------------------------------------------------------------------
Christopher C Stacy <······@world.std.com> writes:
> Robert Posey <·····@raytheon.com> writes:
> > Does any know a LISP system that works on embedded processors?
>
> Minima actually referred to the software part; I have forgotten the model
> designation for of its particular pizza box. I can't even remember if
> it included a disk and paging, or just a boat-load of memory.
> It had multiple Ethernet ports to facilitate redundant network-based
> applications. (Someone will chime in....)
Was that the MB400 multibus system?
Why did AT&T scrap their next-generation switch project?
Is it true that AT&T's pulling out of the Minima stuff is what really
undermined the stability of Symbolics' finances (HP Newquist hints at
this, if I'm remembering right), quickening their eventual bankruptcy?
Also, I understand that one of Lucent's high-end switches does run
lisp (according to Norvig at the last lisp conference). Does anyone
know any more information, such as which model, and what the
hardware/software platforms are? I'd really like to see what it's all
about.
Perspiring minds want to know...
On 19 Jan 2000 11:04:41 -0600, ··@nanofab.utdallas.edu wrote:
> Christopher C Stacy <······@world.std.com> writes:
>
> > Robert Posey <·····@raytheon.com> writes:
> > > Does any know a LISP system that works on embedded processors?
> >
> > Minima actually referred to the software part; I have forgotten the model
> > designation for of its particular pizza box. I can't even remember if
> > it included a disk and paging, or just a boat-load of memory.
> > It had multiple Ethernet ports to facilitate redundant network-based
> > applications. (Someone will chime in....)
>
> Was that the MB400 multibus system?
>
> Why did AT&T scrap their next-generation switch project?
>
> Is it true that AT&T's pulling out of the Minima stuff is what really
> undermined the stability of Symbolics' finances (HP Newquist hints at
> this, if I'm remembering right), quickening their eventual bankruptcy?
>
> Also, I understand that one of Lucent's high-end switches does run
> lisp (according to Norvig at the last lisp conference). Does anyone
> know any more information, such as which model, and what the
> hardware/software platforms are? I'd really like to see what it's all
> about.
>
> Perspiring minds want to know...
Harlequin's 'Hercules' (a real-time LispWorks variant) got the AT&T contract
for the switch project. AFAIK the project was technically successful, but hit
commercial and political difficulties. Hercules ran on VxWorks on SPARC.
Harlequin later applied the VxWorks expertise to the POWER architecture for
NASA.
Various ex-Symbolics ex-Minima people worked at Harlequin but not on Hercules.
__Jason
Date: Wed, 19 Jan 2000 06:34:37 GMT
From: Christopher C Stacy <······@world.std.com>
Robert Posey <·····@raytheon.com> writes:
> Does any know a LISP system that works on embedded processors?
You're probably referring to a different (less powerful) class of
processor and small memory, but since you said "embedded", I'll throw
in some information that is probably only of historical interest to you.
Another historical item:
TI's Lisp chip was developed with some support from the Compact Lisp
Machine program at DARPA. The goal was a four-board system (processor,
two memory, one Nubus-to-Multibus) for some avionics purpose. The
boards were about 4x6 inches with chips on both sides. 40 MHz, 16 MB.
This would be 1987, plus/minus a year.
I don't know what the government did with their prototype or the program
in general -- I'm not a classified person -- but I ended up with our
prototype's memory boards. Ran them (in extender cards) in my Explorer
for a couple of years.
We didn't do anything special in the software for the CLM system, that I
recall. It wasn't uncommon in those days to throw bare hardware at the
military and let them deal with the software themselves, perhaps as part
of another contract.
(TI's defense business has since become part of Raytheon.)
Paul Fuqua
Texas Instruments, Dallas, Texas ··@ti.com
On Tue, 18 Jan 2000 13:29:19 -0600, Robert Posey <·····@raytheon.com>
wrote:
>Dear Gentle Persons,
>
>Does any know a LISP system that works on embedded processors? Is there a
>cheap or free ware version of Embedded LISP. The only version I am aware of
>runs on commercial research and special application robot vendor, but it runs
>on systems with PC Class Processors or above.
Don't forget, you can get cheap embedded processors using the 386EX
and use XLISP under DOS. Just a thought.
Cheers,
Art Ed LeBouthillier
Arthur Ed LeBouthillier wrote:
>
> On Tue, 18 Jan 2000 13:29:19 -0600, Robert Posey <·····@raytheon.com>
> wrote:
>
> >Dear Gentle Persons,
> >
> >Does any know a LISP system that works on embedded processors? Is there a
> >cheap or free ware version of Embedded LISP. The only version I am aware of
> >runs on commercial research and special application robot vendor, but it runs
> >on systems with PC Class Processors or above.
>
> Don't forget, you can get cheap embedded processors using the 386EX
> and use XLISP under DOS. Just a thought.
Those would be fine for most on my applications, but DOS, linux and windows are
a little memory hungry for most simple robots. However, I may well have a LISP
based main controller and use OS free micro controllers for the real time stuff.
Muddy
>
> Cheers,
> Art Ed LeBouthillier