Hi,
does anyone knows about an OS in Lisp or scheme (with a small asm core).
I think I heard or read about it but I have no intention where.
thanx
-- mario
Mario Deilmann wrote:
>
> Hi,
>
> does anyone knows about an OS in Lisp or scheme (with a small asm core).
> I think I heard or read about it but I have no intention where.
Do a Google search on Lisp Machine. Then do one on Lisp OS. There have
been many attempts to construct such a beast after the demise (please
don't argue, I know it will never really be dead) of the Lisp Machines.
Most of these have puttered along for a few months until the main
contributors drift away. These projects usually founder on the
philosophical rocks of:
(a) Do we start with an already existant OS base (for drivers, low-level
memory management, graphics interface, etc.) or roll our own?
(b) Do we write our own Lisp or try to port an existing one?
Most choose an existing OS base and start to write their own Lisp. As
such, they end up with YALI (Yet Another Lisp Implementation) which is
never complete and slowly decays over time.
Others choose trying to build their own OS. As such, they think they
need a Lisp to write it in. So they start to write their own Lisp. As
such, they end up with YALI which doesn't even compile to a real system
and slowly decays over time.
Others try to use available Lisp and OS implementations. They soon
realize that what they have is really not that much better than running
their existing development system which has thoughtfully added most
hooks to the OS. As such, they give up on a Lisp OS and just use that
system.
The one thing that hasn't been tried very often is to build an OS/Lisp
Run Time System, but port an existing compiler to generate code for it.
Since most of the other approaches have failed, it might be worthwhile
to try that.
But it still probably won't be successful. Why? Building an OS is a
BIG job - even using Lisp. It needs lots of grungy, detailed work. And
there's not that many Lispers with that much time to dedicate to the
task. Plus, if you really want a Lisp Machine, you can still get one
(http://www.symbolics.com). In any case, although a real Lisp OS would
be neat, it's just not neat enough for a lot of people to sink a lot of
time into.
faa
In article <·················@qwest.net>,
"Frank A. Adrian" <·······@qwest.net> wrote:
> The one thing that hasn't been tried very often is to build an OS/Lisp
> Run Time System, but port an existing compiler to generate code for it.
> Since most of the other approaches have failed, it might be worthwhile
> to try that.
>
> But it still probably won't be successful. Why? Building an OS is a
> BIG job - even using Lisp. It needs lots of grungy, detailed work. And
> there's not that many Lispers with that much time to dedicate to the
> task. Plus, if you really want a Lisp Machine, you can still get one
> (http://www.symbolics.com). In any case, although a real Lisp OS would
> be neat, it's just not neat enough for a lot of people to sink a lot of
> time into.
More importantly, there aren't that many people who would want to use it
because they lack the understanding or willingness to understand and
appreciate it: "A symbolic *what* ? *noo* *way* - *I* *want* *my*
*windoze* *box* !" ;-)
I remember a program on Discovery Channel, where they showed the design
of a kind of steam engine by Archimedes (?) and the anecdote went that
the inventor demonstrated it to the Roman emperor while enthusiastically
talking about the potential of steam power and the emperor was said to
have responded there was no need to replace the slaves in the empire.
So, the steam powered engine had to wait some 1700 odd years to be
reinvented by James Watt to be deployed and eventually used.
One could certainly argue that the situation with Lisp machines (in the
broader sense) is not unlike that anecdote, whether it had actually
taken place or not.
rgds
BK
--
bk <·········@yahoo.com>
ATTENTION! Email to this address *must* contain "USENET" in the subject line
otherwise it will be considered spam, automatically deleted and never reach me.
Here's TUNES's answers to these questions. (I explained the basic
ideas of TUNES in another post in this thread.)
This is all IIRC because I just officially joined the TUNES project
yesterday and I am a newbie to the project. (But, grepping through the
Summary buffer in Gnus (to mix metaphors), it looks like I'm the only
member subscribed to c.l.l....)
"Frank A. Adrian" <·······@qwest.net> writes:
> Mario Deilmann wrote:
> >
> > Hi,
> >
> > does anyone knows about an OS in Lisp or scheme (with a small asm core).
> > I think I heard or read about it but I have no intention where.
>
> Do a Google search on Lisp Machine. Then do one on Lisp OS. There have
> been many attempts to construct such a beast after the demise (please
> don't argue, I know it will never really be dead) of the Lisp Machines.
> Most of these have puttered along for a few months until the main
> contributors drift away.
TUNES has been around for several years, in design phase.
> These projects usually founder on the
> philosophical rocks of:
>
> (a) Do we start with an already existant OS base (for drivers, low-level
> memory management, graphics interface, etc.) or roll our own?
>
TUNES will initially be a user-level program on any GNU/Linux system
(don't know about other Unices), similar to Squeak in some respects
(one of the project members worked on Squeak, IIRC). Later, there
might be a kernel that specifically provides facilities for, eg,
persistance (TUNES has no file system, only persistance).
> (b) Do we write our own Lisp or try to port an existing one?
>
NIL. (I mean that they choose to write a new implementation of Lisp,
not that they refuse to answer this question :)).
> Most choose an existing OS base and start to write their own Lisp. As
> such, they end up with YALI (Yet Another Lisp Implementation) which is
> never complete and slowly decays over time.
>
I'm not sure if TUNES can be reasonably placed in this category. TUNES
is reminiscent of Squeak, where the software happens to run over an
existing OS but is a wholly seperate environment (and might have a
dedicated kernel in the future). It is much more than just YALI -- I
would describe it as an OS project that is writing an emulator for the
OS before the OS itself.
> Others choose trying to build their own OS. As such, they think they
> need a Lisp to write it in. So they start to write their own Lisp. As
> such, they end up with YALI which doesn't even compile to a real system
> and slowly decays over time.
>
They're not starting out by building their own OS. They might write a
kernel specifically for TUNES, which would be useful since TUNES is so
different from other systems (uses only persistance, for example).
> Others try to use available Lisp and OS implementations. They soon
> realize that what they have is really not that much better than running
> their existing development system which has thoughtfully added most
> hooks to the OS. As such, they give up on a Lisp OS and just use that
> system.
>
I agree that this approach is mostly useless. That's probably why
TUNES doesn't use this method of creating a ``new OS''.
> The one thing that hasn't been tried very often is to build an OS/Lisp
> Run Time System, but port an existing compiler to generate code for it.
> Since most of the other approaches have failed, it might be worthwhile
> to try that.
>
I don't know whether TUNES uses any part of this approach or not.
> But it still probably won't be successful. Why? Building an OS is a
> BIG job - even using Lisp. It needs lots of grungy, detailed work. And
> there's not that many Lispers with that much time to dedicate to the
I don't think TUNES is really centered around a Lisp OS, it's more
like a TheRightThingOS. It just happens to use Lisp for everything.
> task. Plus, if you really want a Lisp Machine, you can still get one
> (http://www.symbolics.com).
I *really* would like to obtain a Lisp Machine. However,
www.symbolics.com seems to be down a lot, and they don't seem to sell
any products there (maybe I looked in the wrong place). Besides, a
Lisp machine would probably be extremely expensive.
> In any case, although a real Lisp OS would
> be neat, it's just not neat enough for a lot of people to sink a lot of
> time into.
>
> faa
--
BPT <···@tunes.org> /"\ ASCII Ribbon Campaign
backronym for Linux: \ / No HTML or RTF in mail
Linux Is Not Unix X No MS-Word in mail
Meme plague ;) ---------> / \ Respect Open Standards
BPT <···@tunes.org> writes:
>
> > task. Plus, if you really want a Lisp Machine, you can still get one
> > (http://www.symbolics.com).
> I *really* would like to obtain a Lisp Machine. However,
> www.symbolics.com seems to be down a lot, and they don't seem to sell
> any products there (maybe I looked in the wrong place). Besides, a
> Lisp machine would probably be extremely expensive.
Well I guess that's the case. The LispOS from Symbolics is called
OpenGenera and just works on TrueUnix (DECS now COMPAQS Unix) on Alpha
Machines. You can buy OpenGenera from Symbolics but you too need an
Alpha WorkStation. Alternativly you may visit
http://www.abstractscience.freeserve.co.uk/symbolics/
and see if Peter Paine can sell you a "used" LispMachine.
Regards
Friedrich
"Mario Deilmann" <······@hotmail.com> writes:
> does anyone knows about an OS in Lisp or scheme (with a small asm
> core). I think I heard or read about it but I have no intention
> where.
The normally-obligatory response is Symbolics Genera, which ran on
either their own hardware, or, more recently, on Digital Alphas atop
Digital Unix. [What with the Compaq takeover, I will feign ignorance
as to what naming should be used now; it's probably only of academic
interest, but that's pretty much what this request is about...]
Several more recent projects have been out there to do this; few have
gotten much past "blathering," although:
- MzScheme has been ported to run atop OSKit, though it lacks ability
to access filesystems or TCP, which makes it of limited usefulness;
- SilkOS, Ocelot, NASOS, all fell out of the "LispOS Project;" dunno
if any of them actually works these days;
- FunOS uses CAML as the implementation language to then build a
FunScheme layer that runs the kernel;
- Sting; an OS written in Scheme;
- Vapour is probably the most-appropriately-named one :-)
--
(concatenate 'string "cbbrowne" ·@ntlug.org")
http://vip.hyperusa.com/~cbbrowne/lisposes.html
Rules of the Evil Overlord #226. "I will have a staff of competent
detectives handy. If I learn that someone in a certain village is
plotting against me, I will have them find out who rather than wipe
out the entire village in a preemptive strike."
<http://www.eviloverlord.com/>
You may be interested in the TUNES� project, at
<URL:http://www.tunes.org/>
Paraphrasing the main WWW page, TUNES is a reflective,
object-oriented, Lisp-based operating system that uses persistance in
place of a file system. I am a member, as you might guess from the
From header and my ~/.signature. (Right now I am designing COOL�, a
Lisp-like programming language that hopefully will influence some
aspects of the HLL for TUNES.)
It is a very interesting system, but there is no running code yet. The
goal is not to make a Lisp OS but to make an excellent OS that happens
to use Lisp because Lisp happens to be the best existing language
(IMHO). The development team is very small, and they don't run around
waving banners that proclaim ``HEY! We're writing a Lisp OS!'' (oops,
I guess I just did ;)), and most of the people who find the project's
page stumble upon it whilst googling for reflective programming or
similar, so it has not yet been torn apart by arguments like previous
LispOS projects, notably LispOS itself.
It also helps that they are apparently not rushing to implement it,
they have been in the design stage for a long time now (because they
basically want to build TheRightThingOS). Recently, they decided to
speed up development, so look for a beginning of an implementation
Real Soon Now(TM).
I am just a newbie to the project, so I hope I have not mis-quoted
anything here. There is *no code* yet, but maybe in 3-5 years there
will be a fairly complete system. So go back to some Unix if you have
unreasonable expectations for an OS such as, for example, expecting it
to exist :).
Footnotes:
� A recursive acronym for ``TUNES is a Useful, Nevertheless
Expedient, System''.
� ``Creatively Object-Oriented Lisp''. So called because it has a
very interesting and ISTM different object system than any other
OO system that I have worked extensively in, namely, Python (which
is the second-best language behind Lisp, IMHO) and C++ (yuck!). I
began designing a while ago, and actually planned to design an OS
for it one day (called KOALA or WALRUS depending on the phase of
the moon), until I found TUNES.
--
BPT <···@tunes.org> /"\ ASCII Ribbon Campaign
backronym for Linux: \ / No HTML or RTF in mail
Linux Is Not Unix X No MS-Word in mail
Meme plague ;) ---------> / \ Respect Open Standards