From: Pascal J. Bourguignon on
Tamas K Papp <tkpapp(a)gmail.com> writes:

> On Tue, 29 Dec 2009 12:48:22 +0200, Victor wrote:
>
>> On Tue, 29 Dec 2009 12:31:14 +0200, Tamas K Papp <tkpapp(a)gmail.com>
>> wrote:
>>
>>> On Tue, 29 Dec 2009 09:03:00 +0000, Alan Mackenzie wrote:
>>>
>>>> gavino <gavcomedy(a)gmail.com> wrote:
>>>>> why is there not a lisp pc for under $300?
>>>>
>>>> There once were lisp machines, in the 1980s. Look up "lisp machine"
>>>> in wikipedia for some interesting history.
>>>
>>> I have a Dell laptop. It runs Lisp. That makes it a Lisp PC.
>>
>> Tamas, AFAIU Lisp Machines were not only merely able to run Lisp
>> compiler/interpreter but substantial parts of the operating system and
>> related system software was written in Lisp, creating an environment for
>> convenient development in Lisp.
>
> Yes, I know what LM were. But I fail to see the big benefit. Of
> course it is "nice" if the OS is written in the same language as an
> application, but as long as they can communicate, it does not matter
> much. Convenience surely does not depend on that. Would my
> SBCL+SLIME+Emacs environment be "more convenient" if it was ticking on
> top of a LM? Why would I care?

Because you are a programmer. It's not the ABI that matter, but the
API, and the possibility you'd have to modify it or its
implementation. You could change your OS, even on the fly,
programming in Lisp.


>> And that's if we put specific details in their hardware design to
>> improve Lisp interpreter performance.
>
> Or---and I know that this sounds crazy---you can just take advantage
> of off-the-shelf hardware, especially the low price and ever improving
> performance, and maybe compile (maybe JIT) your code, so that you don't
> need a hardware interpreter. Just a wild idea.
>
>> Modern personal computers are able to run Lisp code, but PCs are not
>> truly "Lisp Machines" as far as I see.
>
> I think that the distinction is academic. Modern CPUs have so many
> layers anyway that I don't really care if the topmost layer exposed to
> the software is some variant of Lisp or not. If we had a "Lisp machine"
> today, it would most likely be a "microcode machine" that implements
> some Lisp. And then we would have gavino trolling about how it is not
> a LM.

Indeed. Just take a normal PC, and write a Lisp VM kernel. Boot it,
and voil�, Instant Lisp Machine!


> I have noticed that some people are nostalgic about Lisp Machines, and
> I appreciate the sentiment, but to me they look like nothing but a
> glorious dead end. And incidentally, the failure of the Lisp Machine
> market provided another opportunity for people to write Lisp off.

One advantage of a Lisp Machine is that programs written in C would
run slower on it than programs written in Lisp.


--
__Pascal Bourguignon__ http://www.informatimago.com/
From: Alan Mackenzie on
Tamas K Papp <tkpapp(a)gmail.com> wrote:
> On Tue, 29 Dec 2009 12:48:22 +0200, Victor wrote:

>> On Tue, 29 Dec 2009 12:31:14 +0200, Tamas K Papp <tkpapp(a)gmail.com>
>> wrote:

>>> On Tue, 29 Dec 2009 09:03:00 +0000, Alan Mackenzie wrote:

>>>> gavino <gavcomedy(a)gmail.com> wrote:
>>>>> why is there not a lisp pc for under $300?

>>>> There once were lisp machines, in the 1980s. Look up "lisp machine"
>>>> in wikipedia for some interesting history.

>>> I have a Dell laptop. It runs Lisp. That makes it a Lisp PC.

>> Tamas, AFAIU Lisp Machines were not only merely able to run Lisp
>> compiler/interpreter but substantial parts of the operating system and
>> related system software was written in Lisp, creating an environment for
>> convenient development in Lisp.

> Yes, I know what LM were. But I fail to see the big benefit.

You've presumably never used (or played on) one.

> Of course it is "nice" if the OS is written in the same language as an
> application, but as long as they can communicate, it does not matter
> much.

Not only was the Lisp Machine's OS written in lisp, its (microcoded)
object code embodied lisp. For example, the "call" instruction just
pushed the address of a function onto a stack, the actual call being done
after the last argument has been evaluated (the opcode for this
evaluation itself containing a "last operand" field).

> Convenience surely does not depend on that. Would my SBCL+SLIME+Emacs
> environment be "more convenient" if it was ticking on top of a LM?

Yes. Perhaps not for any single great feature, more lots of little
things which are annoyances on Windows or GNU simply just worked on a LM.
Think of doing everything in Emacs, only more so.

> Why would I care?

>> And that's if we put specific details in their hardware design to
>> improve Lisp interpreter performance.

More Lisp object code, though there was of course an interpreter too.

> Or---and I know that this sounds crazy---you can just take advantage
> of off-the-shelf hardware, especially the low price and ever improving
> performance, and maybe compile (maybe JIT) your code, so that you don't
> need a hardware interpreter. Just a wild idea.

>> Modern personal computers are able to run Lisp code, but PCs are not
>> truly "Lisp Machines" as far as I see.

> I think that the distinction is academic. Modern CPUs have so many
> layers anyway that I don't really care if the topmost layer exposed to
> the software is some variant of Lisp or not. If we had a "Lisp machine"
> today, it would most likely be a "microcode machine" that implements
> some Lisp. And then we would have gavino trolling about how it is not
> a LM.

> I have noticed that some people are nostalgic about Lisp Machines, and
> I appreciate the sentiment, but to me they look like nothing but a
> glorious dead end. And incidentally, the failure of the Lisp Machine
> market provided another opportunity for people to write Lisp off.

Yes indeed. The Intel PC (language: C) won over the Lisp Machine
(language: lisp). My own opinion (for what it's worth) is that it takes
a much more capable programmer to master Lisp than C, and in the late
1980s there weren't enough Lispers around to make a PC priced LM
economic.

> Cheers,

> Tamas

--
Alan Mackenzie (Nuremberg, Germany).

From: Citizen Jimserac on
On Dec 29, 5:31 am, Tamas K Papp <tkp...(a)gmail.com> wrote:
> On Tue, 29 Dec 2009 09:03:00 +0000, Alan Mackenzie wrote:
> > gavino <gavcom...(a)gmail.com> wrote:
> >> why is there not a lisp pc for under $300?
>
> > There once were lisp machines, in the 1980s.  Look up "lisp machine" in
> > wikipedia for some interesting history.
>
> I have a Dell laptop.  It runs Lisp.  That makes it a Lisp PC.

NOT EVEN CLOSE!!

Gavino has a point and implies something, I think, which I will try to
extrapolate on.

In the world of the future as we envisioned it in the 60's and early
70's, there would be hotels on the moon, intelligent HAL like
computers and like science fiction staple advancements.

In reality, while in the realm of computer hardware, such advances
occured on time or even ahead of "schedule" many of the other things
are not even close to happening. Why not?

Some will say because the problems were far harder than originally
understood. That is true up to a point, I agree.

But there is much more - I believe the misappropriation and
misapplication of government funds, the politicization of research,
the dangerously subversive and unproductive influence of the lobbyists
and the deliberate quelling of the American futuristic imagination and
desire after the Vietnam war are some of the more real factors
involved in the inequality between expectation and reality that
actually happened. Likewise, if the organizational difficulties
endured by Dr. Schelter were any example, the entire foundations of
research and development infrastructure in this country need to be
seriously rebuilt if we are ever to hope to return to the post
Sputnik era of high developement attainments over relatively short
periods of time. The obstructions ... and obstructors are many.

Observe the appalling mismanagement at NASA and the obvious directions
of the future when Bert Rutan was able to accomplish, on his own,
something formerly done only by governments. The prize was a measly
million dollars. One wonders what he could have accomplished with $20
million or 50 million - large sums of money but a pittance to the
budget.
I think it was only recently that the software involved in the LEM
guidance was released to the public domain, something that properly
should have been done decades ago. AND, TONS of code and research
remain locked up by NASA and subcontractors that, although
declassified, requires $$$ to obtain. Inexcusable since much of it
was written on the taxpayers own tab.


Observe the appalling waste of money on cancer "research" - yes some
breakthroughs over the years, even major ones, but how much of that
money is being wasted, one wonders, on barking up the wrong tree in
the wrong forest for the wrong reasons. See "The Secret History of
the War on Cancer" by Davis for details.

Think of the impact that freely available Lisp Machine and Symbolics
type software would have on these activities!

Citizen Jimserac
From: Tamas K Papp on
On Tue, 29 Dec 2009 16:50:46 +0000, Alan Mackenzie wrote:

> Tamas K Papp <tkpapp(a)gmail.com> wrote:
>> I have noticed that some people are nostalgic about Lisp Machines, and
>> I appreciate the sentiment, but to me they look like nothing but a
>> glorious dead end. And incidentally, the failure of the Lisp Machine
>> market provided another opportunity for people to write Lisp off.
>
> Yes indeed. The Intel PC (language: C) won over the Lisp Machine
> (language: lisp). My own opinion (for what it's worth) is that it takes
> a much more capable programmer to master Lisp than C, and in the late
> 1980s there weren't enough Lispers around to make a PC priced LM
> economic.

Sorry, but this rationalization sounds like sour grapes.

The Intel PC is not a "C machine": various CPU designs are equally
close to C, the Intel PC is not special. The architecture won because
it was cheap and _general_. Generality is the key. You can jolly
well program Lisp on an Intel PC, but what if you want to use a
significantly different language on a Lisp machine? So unless you are
100% sure you want Lisp, you will not risk buying an expensive
specialized machine.

Having specialized hardware is always very expensive. Given the high
price of competent labor and the labor intensiveness of hardware
design & development, it is rarely worth it. There are of course
exceptions, and flexible options (eg FPGAs) make specialized hardware
viable in some limited scenarios.

Tamas
From: Tamas K Papp on
On Tue, 29 Dec 2009 09:59:25 -0800, Citizen Jimserac wrote:

> Gavino has a point and implies something, I think, which I will try to
> extrapolate on.

Gavino doesn't even understand his own questions, even if they
accidentally make sense. Him having a point is a rather far-fetched
idea.

> Observe the appalling waste of money on cancer "research" - yes some
> breakthroughs over the years, even major ones, but how much of that
> money is being wasted, one wonders, on barking up the wrong tree in the
> wrong forest for the wrong reasons. See "The Secret History of the War
> on Cancer" by Davis for details.
>
> Think of the impact that freely available Lisp Machine and Symbolics
> type software would have on these activities!

You didn't provide any arguments to substantiate this. I don't deny
the possibility that Lisp machines are a neat thing to have, but to
argue that they provide some magical benefit that revolutionizes
cancer research etc is nonsensical.

Here is a puzzle for you: if Lisp machines are so great, why don't we
have any nowadays? It is not that they are prohibitively expensive to
make. As Pascal remarked above, you can write a Lisp VM on an Intel
PC and have your very own Lisp Machine, with many (if not all) of the
claimed benefits. But no one is doing this, so the benefits must be
quite small.

Cheers,

Tamas