Hello guys
I heard somebody (Andrei Alexandrescu) somewhere
(comp.lang.c++.moderated) saying something (precise quote at
end of post) about major crash of lisp-based shops in 80's.
I am too young to know about 80's. I was thinking IT is in
constant expansion. I wonder what a catastrophe could bring
large group of IT companies to crash? Was there something
wrong with language itself? Or was it marketing games? Or
was it some global change? You guys should know something
about it.
Thanks for your answer.
/Robert
****** begin quote (Andrei Alexandrescu) *******
"Thant Tessman" <·····@acm.org> wrote
> Matt Seitz wrote:
> > [...] Otherwise, at best it merely serves to convince current
> > Lisp programmers to not try C++.
>
> Believe me, they don't need any convincing.
I guess all guys who got laid off after the major crash and
burn of all
LISP-based shops in the eighties didn't, either. Besides, of
course you
don't need any convincing to stay with a language that is
still hopelessly
slow after a 10,000 fold increase in computing speed.
****** end quote *******
Subject: Re: Acolyte of the Week #20: Why so much undefined
behavior?
Message-ID: <··············@ID-14036.news.dfncis.de>
Robert Vazan wrote:
>
> Hello guys
>
> I heard somebody (Andrei Alexandrescu) somewhere
> (comp.lang.c++.moderated) saying something (precise quote at
> end of post) about major crash of lisp-based shops in 80's.
> I am too young to know about 80's. I was thinking IT is in
> constant expansion. I wonder what a catastrophe could bring
> large group of IT companies to crash? Was there something
> wrong with language itself? Or was it marketing games? Or
> was it some global change? You guys should know something
> about it.
That would have been the "AI Winter". AI and Lisp shops were in
roughly the roughly same position as the dot-com companies of more
recent note, and fell from grace similarly. The markets and venture
capitalists abandoned them when AI failed to live up to their
unrealistic expectations.
Many people on this list have some rather bitter memories of those dark
days, so I'm not sure how forthcoming they'll be.
Michael Parker <·······@pdq.net> writes:
> Robert Vazan wrote:
> >
> > Hello guys
> >
> > I heard somebody (Andrei Alexandrescu) somewhere
> > (comp.lang.c++.moderated) saying something (precise quote at
> > end of post) about major crash of lisp-based shops in 80's.
> > I am too young to know about 80's. I was thinking IT is in
> > constant expansion. I wonder what a catastrophe could bring
> > large group of IT companies to crash? Was there something
> > wrong with language itself? Or was it marketing games? Or
> > was it some global change? You guys should know something
> > about it.
>
> That would have been the "AI Winter". AI and Lisp shops were in
> roughly the roughly same position as the dot-com companies of more
> recent note, and fell from grace similarly. The markets and venture
> capitalists abandoned them when AI failed to live up to their
> unrealistic expectations.
It was worse than that; many large companies sunk billions into AI,
having been given some strong promises of thinking machines, language
recognition, expert systems, etc. Some of the promises were nothing but
hype, and other more realistic goals were nevertheless much harder to
attain than originally thought. It was, as always, a question of money
and return-on-investment - when AI groups started failing to deliver as
promised, they started losing their funding and became the pariah of
those companies. Lisp was one of the major implementation langauages
of these efforts, and so by association lisp's name was tainted by the
"crash". I don't know if there was any one event that started or
ended the AI winter, but having gone through it I would estimate from
about 1989 through 1993 were the bleakest years, with lisp finally getting
a new foothold in 95 or 96 after most of us went to the web.
Not all lisp-based shops crashed or burned; some of them went underground.
Some of them still are, and don't mention that they get their job done
quickly _because_ there is lisp inside. In fact, some keep this fact
quiet as a competitive advantage.
> Many people on this list have some rather bitter memories of those dark
> days, so I'm not sure how forthcoming they'll be.
Some people have been badly burned. At my company, we were hand-to-mouth
for a while (though still regularly profitable, since we had/have good
business practices) and it was rough for a while. Also, I think that the
colapse of the lisp-machine market caused rough going for those of us
who were in the lisp-on-general-purpose-hardware business. Though I
don't view lisp machine decline as a product or cause of the AI winter,
but as a result of the hard reality of limited man-years to devote to
lisp-machine development, as compared to the orders of magnitude more
manpower that gp-hardware vendors could devote to development and scaling.
I for one am forthcoming because I believe the worst is past us, and
lisp is once again on the rise, slowly and steadily.
--
Duane Rettig Franz Inc. http://www.franz.com/ (www)
1995 University Ave Suite 275 Berkeley, CA 94704
Phone: (510) 548-3600; FAX: (510) 548-8253 ·····@Franz.COM (internet)
Duane Rettig <·····@franz.com> writes:
> Michael Parker <·······@pdq.net> writes:
> > That would have been the "AI Winter". AI and Lisp shops were in
> > roughly the roughly same position as the dot-com companies of more
> > recent note, and fell from grace similarly. The markets and venture
> > capitalists abandoned them when AI failed to live up to their
> > unrealistic expectations.
>
> It was worse than that; many large companies sunk billions into AI,
I dunno if it was worse in any reasonable sense: a huge amount of
money has been wiped off stockmarkets recently, and probably a fair
amount of that can be put down to ludicrous expectations of new
technology leading to completely unrealistic valuations of companies -
not just small dot-coms, either.
--tim
Tim Bradshaw <···@tfeb.org> writes:
> Duane Rettig <·····@franz.com> writes:
>
> > Michael Parker <·······@pdq.net> writes:
>
> > > That would have been the "AI Winter". AI and Lisp shops were in
> > > roughly the roughly same position as the dot-com companies of more
> > > recent note, and fell from grace similarly. The markets and venture
> > > capitalists abandoned them when AI failed to live up to their
> > > unrealistic expectations.
> >
> > It was worse than that; many large companies sunk billions into AI,
>
> I dunno if it was worse in any reasonable sense: a huge amount of
> money has been wiped off stockmarkets recently, and probably a fair
> amount of that can be put down to ludicrous expectations of new
> technology leading to completely unrealistic valuations of companies -
> not just small dot-coms, either.
Precisely, and already the term "dot-com" has taken on a new, bad
connotation, because of the strong association with the hype. Also,
fewer companies are announcing themselves as "pre-IPO", for the same
reason. I'm sure venture-capitalists will think twice about sinking
their money into dot-coms which have no prospects of making a profit
in a reasonable time (at least, I hope they're smart enough to be shy
like this). Unfortunately, this will also make it harder for
legitimate dot-coms with real workable business plans to get funding
for these efforts.
The difference, however, was that while waiting for a profit, dot-coms
do provide benefits (unlike a partially completed AI project, which
usually can't be used until it is operational). Thus, though venture
capitalists will walk away from a small ipo that will obviously never
make a profit, a large company with a website will continue to fund
such an effort (although possibly on a smaller scale), count it as
marketing loss-leader, and reap the sales/marketing benefits it has
to offer. In this way, a middle-to-upper manager doesn't put his
job on the line, because "internet" and "web" have not become bad words,
but only "pre-ipo" and "dot-com", neither of which his company has
anything to do with.
--
Duane Rettig Franz Inc. http://www.franz.com/ (www)
1995 University Ave Suite 275 Berkeley, CA 94704
Phone: (510) 548-3600; FAX: (510) 548-8253 ·····@Franz.COM (internet)
Duane Rettig <·····@franz.com> writes:
>
> The difference, however, was that while waiting for a profit, dot-coms
> do provide benefits (unlike a partially completed AI project, which
> usually can't be used until it is operational). Thus, though venture
> capitalists will walk away from a small ipo that will obviously never
> make a profit, a large company with a website will continue to fund
> such an effort (although possibly on a smaller scale), count it as
> marketing loss-leader, and reap the sales/marketing benefits it has
> to offer. In this way, a middle-to-upper manager doesn't put his
> job on the line, because "internet" and "web" have not become bad words,
> but only "pre-ipo" and "dot-com", neither of which his company has
> anything to do with.
>
Also, I think, there's a difference in that lots of dot-coms actually
made something that basically worked -- even if badly -- although it
was (or is now) clear that they couldn't make money doing this. After
all, a lot of it just comes down to selling x over the net (books,
plane tickets &c), and it's kind of obvious it can be done. Whereas
AI promised solutions to all sorts of open research problems, which
solutions it turned out it could not deliver at all. Failing because
you can't make a profit is not so bad as failing because you claimed
you could do things which you can't, actually do.
(I think I've maybe just restated what you said , excetpy more
obscurely, damn).
--tim
In article <···············@tfeb.org>, Tim Bradshaw <···@tfeb.org> wrote:
>Duane Rettig <·····@franz.com> writes:
>
>>
>> The difference, however, was that while waiting for a profit, dot-coms
>> do provide benefits (unlike a partially completed AI project, which
>> usually can't be used until it is operational). Thus, though venture
>> capitalists will walk away from a small ipo that will obviously never
>> make a profit, a large company with a website will continue to fund
>> such an effort (although possibly on a smaller scale), count it as
>> marketing loss-leader, and reap the sales/marketing benefits it has
>> to offer. In this way, a middle-to-upper manager doesn't put his
>> job on the line, because "internet" and "web" have not become bad words,
>> but only "pre-ipo" and "dot-com", neither of which his company has
>> anything to do with.
>>
>
>Also, I think, there's a difference in that lots of dot-coms actually
>made something that basically worked -- even if badly -- although it
>was (or is now) clear that they couldn't make money doing this. After
>all, a lot of it just comes down to selling x over the net (books,
>plane tickets &c), and it's kind of obvious it can be done. Whereas
>AI promised solutions to all sorts of open research problems, which
>solutions it turned out it could not deliver at all. Failing because
>you can't make a profit is not so bad as failing because you claimed
>you could do things which you can't, actually do.
I'm not sure the above analyses are correct. What many dotcoms produced
in terms of web sites (or tools for them) was not significantly more useful
than your typical prototype AI project. The difference was more that Web
stuff was radiating into an essentially new space, whereas the AI folks had
to compete with existing applications folks who had 20-30 years of really
good experience in pushing back. (In some ways, both foundered not so
much on inability to deliver useful stuff as inability to find people willing to
_pay_ for things they could in fact deliver.)
In addition, the AI bubble was never nearly as big in dollar terms as it was
in hype. Tens of millions, maybe hundreds of millions at the outside (I don't
think DARPA et al ever got over about $100M a year).
I think, however, that Duane is right about this being potentially the
right time for a lisp resurgence, not only thanks to Moore's law but also
to a decade of people coming to appreciate the notion of systems that
don't always work and that fail gracefully.
paul
From: James Hague
Subject: Re: a bit of history of lisp
Date:
Message-ID: <3b151aae_3@newsfeeds>
Paul Wallich wrote:
>
> I think, however, that Duane is right about this being potentially the
> right time for a lisp resurgence, not only thanks to Moore's law but also
> to a decade of people coming to appreciate the notion of systems that
> don't always work and that fail gracefully.
I agree with you, though I can also see this being a resurgence of the
"Where can I get a good, inexpensive Lisp implementation?" question. Unlike
people coming to Python, for example, there are fewer good answers.
James
-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----== Over 80,000 Newsgroups - 16 Different Servers! =-----
"James Hague" <···········@volition-inc.com> writes:
> Paul Wallich wrote:
> >
> > I think, however, that Duane is right about this being potentially the
> > right time for a lisp resurgence, not only thanks to Moore's law but also
> > to a decade of people coming to appreciate the notion of systems that
> > don't always work and that fail gracefully.
>
> I agree with you, though I can also see this being a resurgence of the
> "Where can I get a good, inexpensive Lisp implementation?" question. Unlike
> people coming to Python, for example, there are fewer good answers.
What do you mean? There are many good and inexpensive (gratis) CL
implementations. All implementing a far more useable and powerful
language than tlotbafir. For tlotbafir you have essentially two
gratis implementations (the original one and the Java based one). I
believe you are mistaken.
Cheers
--
Marco Antoniotti ========================================================
NYU Courant Bioinformatics Group tel. +1 - 212 - 998 3488
719 Broadway 12th Floor fax +1 - 212 - 995 4122
New York, NY 10003, USA http://bioinformatics.cat.nyu.edu
"Hello New York! We'll do what we can!"
Bill Murray in `Ghostbusters'.
Marco Antoniotti <·······@cs.nyu.edu> writes:
> implementations. All implementing a far more useable and powerful
> language than tlotbafir. For tlotbafir you have essentially two
Shouldn't that be "tlotdafir" or maybe "t7r" ;-) Then again it is your
coinage so there is some room for poetic license.
-Eric "Garsh I shouldna done no spellin flame" Dahlman
From: James Hague
Subject: Re: a bit of history of lisp
Date:
Message-ID: <3b1542d9$1_4@newsfeeds>
Marco Antoniotti wrote:
>
> What do you mean? There are many good and inexpensive (gratis) CL
> implementations.
> [...]
> I believe you are mistaken.
Perhaps I should rephrase my original comments. Yes, you can get free
Common Lisp implementations for the usual desktop operating systems, but
there is a very large gap between the inexpensive and the expensive. Lisp
users just hate it when someone calls Lisp slow. But in all honestly, to
get good compilers you have to pay a large sum of money (over $2000 for the
base version of Allegro CL). That's a different story than with, say, C++
compilers, where the free gcc holds its own, and you can get commercial
offerings from $100-$500.
I am not saying that good technology should be free or that commercial Lisps
should be priced similarly to C++ compilers. I am only pointing out that
newcomers to the language will--and do--ask about such things. That's a
hard issue to dodge. (BTW, the best answer for Windows is "Corman Common
Lisp." Very nice, though ACL generates significantly better code. If Roger
worked on that, Franz would start getting a bit worried, I think.)
James
-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----== Over 80,000 Newsgroups - 16 Different Servers! =-----
"James Hague" <···········@volition-inc.com> writes:
> Marco Antoniotti wrote:
> >
> > What do you mean? There are many good and inexpensive (gratis) CL
> > implementations.
> > [...]
> > I believe you are mistaken.
>
> Perhaps I should rephrase my original comments. Yes, you can get free
> Common Lisp implementations for the usual desktop operating systems, but
> there is a very large gap between the inexpensive and the expensive. Lisp
> users just hate it when someone calls Lisp slow. But in all honestly, to
> get good compilers you have to pay a large sum of money (over $2000 for the
> base version of Allegro CL). That's a different story than with, say, C++
> compilers, where the free gcc holds its own, and you can get commercial
> offerings from $100-$500.
Simply False. Check out http://www.cons.org.
> I am not saying that good technology should be free or that commercial Lisps
> should be priced similarly to C++ compilers. I am only pointing out that
> newcomers to the language will--and do--ask about such things. That's a
> hard issue to dodge. (BTW, the best answer for Windows is "Corman Common
> Lisp." Very nice, though ACL generates significantly better code. If Roger
> worked on that, Franz would start getting a bit worried, I think.)
See my previous comment.
Cheers
--
Marco Antoniotti ========================================================
NYU Courant Bioinformatics Group tel. +1 - 212 - 998 3488
719 Broadway 12th Floor fax +1 - 212 - 995 4122
New York, NY 10003, USA http://bioinformatics.cat.nyu.edu
"Hello New York! We'll do what we can!"
Bill Murray in `Ghostbusters'.
"James Hague" <···········@volition-inc.com> writes:
> Yes, you can get free Common Lisp implementations for the usual
> desktop operating systems, but there is a very large gap between the
> inexpensive and the expensive. Lisp users just hate it when someone
> calls Lisp slow. But in all honestly, to get good compilers you
> have to pay a large sum of money (over $2000 for the base version of
> Allegro CL). That's a different story than with, say, C++
> compilers, where the free gcc holds its own, and you can get
> commercial offerings from $100-$500.
Xanalys LispWorks Professional with IDE is just a little bit above in
price and *very* much above in all other aspects. And it also includes
CLIM.
And free Clisp or CMUCL hold their own no worse than gcc.
--
Janis Dzerins
If million people say a stupid thing it's still a stupid thing.
From: James Hague
Subject: Re: a bit of history of lisp
Date:
Message-ID: <3b1651d3_3@newsfeeds>
Janis Dzerins wrote:
>
> Xanalys LispWorks Professional with IDE is just a little bit above in
> price and *very* much above in all other aspects. And it also includes
> CLIM.
>
> And free Clisp or CMUCL hold their own no worse than gcc.
There is too much explaining away of this, methinks. Let's see some
benchmarks comparing CMUCL, Clisp, LispWorks, and ACL. I have been under
the impression that (A) CMUCL is not available for Windows, and (B) Clisp is
at least an order of magnitude slower than the slowest of the machine code
generating Lisps available for Windows.
James
-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----== Over 80,000 Newsgroups - 16 Different Servers! =-----
> (B) Clisp is
> at least an order of magnitude slower than the slowest of the machine code
> generating Lisps available for Windows.
On Linux, Clisp is considerably slower than Lispworks even with lw
running uncompiled code, clisp performance also seems to seriously
degrade when lots of consing is going/has gone on. I've not measured
comprehensively, but the difference is dramatic. I think its
comfortably a factor of 5 given a substantial workload. On the other
hand, Clisp is small and portable- it works wonderfully as a
replacement for bash scripts, particularly on Windows w/ Cygwin; and
has fewer bumps and wierdnesses than Python (IMO).
Gregm
Greg Menke <··········@mindspring.com> writes:
> Clisp is small and portable- it works wonderfully as a
> replacement for bash scripts, particularly on Windows w/ Cygwin; and
> has fewer bumps and wierdnesses than Python (IMO).
With a small amount of configuration, you can use it (on Linux) as a
login shell. In combination with the readline functionality, it's
quite comfortable as a replacement for a shell.
Regards,
Peter
Peter Wood <··········@worldonline.dk> wrote in message news:<··············@localhost.localdomain>...
> Greg Menke <··········@mindspring.com> writes:
>
> > Clisp is small and portable- it works wonderfully as a
> > replacement for bash scripts, particularly on Windows w/ Cygwin; and
> > has fewer bumps and wierdnesses than Python (IMO).
>
> With a small amount of configuration, you can use it (on Linux) as a
> login shell. In combination with the readline functionality, it's
> quite comfortable as a replacement for a shell.
Peter,
I can't be the only person here who'd be very appreciative if you'd
say more about the small amount of configuration required to run
CLISP as a Linux shell. That sounds great since I'm using CLISP on
Linux to learn CL (working through Lamkins online book, which is quite
good, I think).
Thanks,
Kendall Clark
·······@monkeyfist.com (Kendall Clark) writes:
> Peter Wood <··········@worldonline.dk> wrote
> > Greg Menke <··········@mindspring.com> writes:
> >
> > > Clisp is small and portable- it works wonderfully as a
> > > replacement for bash scripts, particularly on Windows w/ Cygwin; and
> > > has fewer bumps and wierdnesses than Python (IMO).
> >
> > With a small amount of configuration, you can use it (on Linux) as a
> > login shell. In combination with the readline functionality, it's
> > quite comfortable as a replacement for a shell.
>
> Peter,
>
> I can't be the only person here who'd be very appreciative if you'd
> say more about the small amount of configuration required to run
> CLISP as a Linux shell. That sounds great since I'm using CLISP on
> Linux to learn CL (working through Lamkins online book, which is quite
> good, I think).
(Another excellent book for beginners is Dave Touretzky's book which
is available for personal (educational) use on his website.)
The following works for me:
*DISCLAIMER: Use at own risk. Exercise care. Make backups of anything
you change.*
On linux, you need to have built your clisp with the
"--with-modules=bindings/linuxlibc6" option, and you should *not* have
compiled it without readline. See the build instructions and Makefile.
STEP 1
Put /usr/bin/clisp in /etc/shells so it looks (something like) this:
# /etc/shells: valid login shells
#/bin/ash
/bin/bash
/bin/sh
/usr/bin/clisp
STEP 2
You can set up your $PATH In /etc/login.defs by modifying the entry
for ENV_PATH to look (something) like this:
ENV_PATH PATH=/bin:/usr/bin:/usr/local/bin:/usr/X11R6/bin
This is so you can run X.
[Once you have your clisp shell set up, you can control the
environment variables (on Linux) via the linuxlibc6 bindings (see
$clisp-src/modules/bindings/linuxlibc6/linux.lisp): setenv, putenv,
getenv, etc]
STEP 3
The "startx" command is a shell script, so we can't use it. As a
temporary fix, I have defunned thus:
(defun startx ()
(execute "/usr/X11R6/bin/xinit"))
Which works for me. Although I haven't investigated all the
consequences of not going via shell's startx. Also I *don't* use
GNOME or KDE which may complicate things for you if you do. I use
fvwm2 and have this in my .xinitrc:
exec fvwm2
STEP 4 (some guru will probably shoot this down in flames)
For conveniently running external programs (ie, I don't want to have
to type (run-program "ls" :arguments '("-lh")) every time, I have set
up a read macro. Put the following in somefile.lisp.
(set-macro-character #\] (get-macro-character #\)))
(set-dispatch-macro-character #\# #\[
#'(lambda (stream char1 char2)
(setf (readtable-case *readtable*) :preserve)
(UNWIND-PROTECT
(LET* ((OUTPUT-STREAM (MAKE-STRING-OUTPUT-STREAM
:ELEMENT-TYPE 'BASE-CHAR))
(STEP1 (FORMAT OUTPUT-STREAM "(RUN-PROGRAM "))
(COMMAND-LINE (READ-DELIMITED-LIST #\] STREAM T))
(COMMAND (FIRST COMMAND-LINE))
(STEP2 (FORMAT OUTPUT-STREAM "\"~A\"
:ARGUMENTS " COMMAND))
(PARAMETERS (REST COMMAND-LINE))
(STEP3 (FORMAT OUTPUT-STREAM "'(")))
(DOLIST (X PARAMETERS
(PROGN (FORMAT OUTPUT-STREAM "))")
(LET ((CLEAN
(GET-OUTPUT-STREAM-STRING
OUTPUT-STREAM)))
(CLOSE OUTPUT-STREAM)
(VALUES
(READ-FROM-STRING CLEAN)))))
(FORMAT OUTPUT-STREAM "\"~A\" " X)))
(SETF (READTABLE-CASE *READTABLE*) :UPCASE))))
You will get (harmless) warnings from the compiler about unused
variables STEP*, if you compile this file. You must keep the upper
case stuff as is. (It could also be all upcase).
STEP 5
It's also a PITA to have to type "#[" and "]" everytime you want to
run a command, so add the following in your $HOME/.inputrc
#for clash: prints the square brackets to run an external command
"\ec": "#[]\C-b"
When you type ESC-c (or META-c) readline will print "#[]" to the
console and put the cursor inside the brackets. If you are running
clisp as your shell, and do this, you will then be able to run
programs (more or less) normally. You will need to escape the dot
in any filenames with a dot in them: #[cat \.xinitrc]
STEP 6
Now try it out-- *Don't* modify your /etc/passwd yet! --
Start up a clisp with "clisp -k full", and load somefile.lisp (the
read-macro) into a CLISP which has readline compiled in.
Hit "ESC c", type "ls -l", and see if it works.
You *don't* have command line completion for external programs inside
the read-macro. So hitting tab will just give you a list of all
clisp's possibilties, which is not much use here. However,
"run-program" does search your path for executables so you don't have
to do #[/bin/ls -l]
Also, I haven't (yet) written lisp versions of "cd" and the other bash
builtins. But this is easy with the libc6 bindings, (chdir etc).
Another convenience thing you can do is stuff like this for often used
commands:
(defmacro ls-lh ()
`#[ls -lh])
Now (ls-lh) reads as (run-program "ls" :arguments '("-lh"))
STEP 7
If you are happy, and think you can live with this setup then you must
do the following:
-Dump a memory image from a clisp with the read macro and any
convenience stuff loaded:
(saveinitmem "lispinit.mem" :quiet t)
The :quiet t suppresses clisp's start up message, which I find
annoying. If you have some function you want to run every time clisp
starts, you can also specify this (see impnotes.html).
-Make a backup of /usr/[local]/lib/clisp. (wherever your's is
installed)
-Rename the "base" directory in /usr/lib/clisp to "orig-base".
-Now rename the "full" directory to "base".
-Replace the memory image in your *new* "base" directory
with the one you dumped.
STEP 8
Change your /etc/passwd to reflect your new shell.
Light 13 candles in a circle round your Linux box, sacrifice a white
rooster, invoke Saint IGNUcious and ...
Login again.
If you start pining for Bash, you can run:
#[bash]
And explicitly source all your configuration files (.profile etc)
--------------------------------------------------------------------
It works for me. Now I just need to work out how to:
get indentation on the command line.
do command redirection.
get name completion for arguments which are filenames.
Piece of cake! (But don't hold your breath)
Have fun,
Peter
What about adding this info to CLiki?
Paolo
--
EncyCMUCLopedia * Extensive collection of CMU Common Lisp documentation
http://cvs2.cons.org:8000/cmucl/doc/EncyCMUCLopedia/
* Paolo Amoroso <·······@mclink.it>
| What about adding this info to CLiki?
done. <URL:http://ww.telent.net/cliki/Lisp%20as%20a%20shell>
I didn't figure out how to stop Cliki from inserting <p>s between the
lines of code in the <pre> blocks, though. anyone who knows?
--
Vebjorn
From: Stig E. Sandoe
Subject: Re: CLISP as Linux shell (was Re: a bit of history of lisp)
Date:
Message-ID: <87vgmfsmt8.fsf@palomba.bananos.org>
Vebjorn Ljosa <·····@ljosa.com> writes:
> * Paolo Amoroso <·······@mclink.it>
> | What about adding this info to CLiki?
>
> done. <URL:http://ww.telent.net/cliki/Lisp%20as%20a%20shell>
This was done yesterday actually, maybe you should check the Cliki
before you add stuff, e.g the Recent Changes section or the
CLISP-page might give hints of any existinging pages.
--
------------------------------------------------------------------
Stig Erik Sandoe ····@ii.uib.no http://www.ii.uib.no/~stig/
From: Stig E. Sandoe
Subject: Re: CLISP as Linux shell (was Re: a bit of history of lisp)
Date:
Message-ID: <87snhjsmss.fsf@palomba.bananos.org>
Vebjorn Ljosa <·····@ljosa.com> writes:
> * Paolo Amoroso <·······@mclink.it>
> | What about adding this info to CLiki?
>
> done. <URL:http://ww.telent.net/cliki/Lisp%20as%20a%20shell>
This was done yesterday actually, maybe you should check the Cliki
before you add stuff, e.g the Recent Changes section or the
CLISP-page might give hints of any existing pages.
--
------------------------------------------------------------------
Stig Erik Sandoe ····@ii.uib.no http://www.ii.uib.no/~stig/
* ····@ii.uib.no (Stig E. Sandoe)
|
| This was done yesterday actually, maybe you should check the Cliki
| before you add stuff, e.g the Recent Changes section or the
| CLISP-page might give hints of any existing pages.
hm, I did...must have been temporarily blind. I removed my page and
linked to yours.
--
Vebjorn Ljosa
Vebjorn Ljosa <·····@ljosa.com> writes:
> * Paolo Amoroso <·······@mclink.it>
> | What about adding this info to CLiki?
>
> done. <URL:http://ww.telent.net/cliki/Lisp%20as%20a%20shell>
>
> I didn't figure out how to stop Cliki from inserting <p>s between the
> lines of code in the <pre> blocks, though. anyone who knows?
Hi
Sam Steingold has put an improved version of the instructions on the
CLISP site
<url:http://clisp.cons.org>
Amongst other things, he points out that CLISP's built in (ext:shell
command) can be called from the CLISP login shell to do stuff like:
[5]> (shell "echo hi")
hi
0
[6]>
You should be able to run shell scripts, one way or the other.
Although if you call (shell) you obviously just get another CLISP.
Regards,
Peter
"James Hague" <···········@volition-inc.com> writes:
> Janis Dzerins wrote:
> >
> > Xanalys LispWorks Professional with IDE is just a little bit above in
> > price and *very* much above in all other aspects. And it also includes
> > CLIM.
> >
> > And free Clisp or CMUCL hold their own no worse than gcc.
>
> There is too much explaining away of this, methinks. Let's see some
> benchmarks comparing CMUCL, Clisp, LispWorks, and ACL. I have been under
> the impression that (A) CMUCL is not available for Windows, and (B) Clisp is
> at least an order of magnitude slower than the slowest of the machine code
> generating Lisps available for Windows.
So what we have at the end is not-so expensive LispWorks. (And there's
also Corman Lisp if you're so into winblows.)
--
Janis Dzerins
If million people say a stupid thing it's still a stupid thing.
"James Hague" <···········@volition-inc.com> writes:
> "Not so expensive" is "$799." Whether or not that's expensive for a
> development environment is up for debate.
What's your charge-out rate? This is probably ~10 hours of it if you
have a salary. It's cheap.
--tim
In article <···············@tfeb.org>, Tim Bradshaw <···@tfeb.org> wrote:
>"James Hague" <···········@volition-inc.com> writes:
>
>> "Not so expensive" is "$799." Whether or not that's expensive for a
>> development environment is up for debate.
>
>What's your charge-out rate? This is probably ~10 hours of it if you
>have a salary. It's cheap.
>
For the person who uses any language professionally, every language
system I have seen for Windows/Mac is cheap. None of them are
more than a few thousand, and if that's more than a few week's
revenues you've got bigger problems than a language system will
solve.
The question is how much is reasonable to charge so that non-professionals
are interested. Franz is definitely out of that price range,
Xanalysis probably, Digitool maybe (the cheapest of the three,
unless you want redistribution rights, in which case it's in
the middle) (and, yes, Digitool is Mac-only).
So, you can get demo Lisp systems on all common desktop OSs.
You can get CMUCL on Linux, and Corman Lisp on Windows, and
AFAIK there's nothing on Macs between the MCL demo and MCL
itself (is Power Lisp still around?)
Whether this is important depends on how you think programmers
are going to get into Lisp. If you think they will be driven
externally, it's no big deal. If you think that a grass-roots
Lisp movement is necessary, it's a problem.
--
David H. Thornley | If you want my opinion, ask.
·····@thornley.net | If you don't, flee.
http://www.thornley.net/~thornley/david/ | O-
David Thornley wrote:
[snip]
>
> The question is how much is reasonable to charge so that non-professionals
> are interested.
Or professionals using other languages who decide to take a look at LISP.
> Franz is definitely out of that price range,
> Xanalysis probably, Digitool maybe (the cheapest of the three,
> unless you want redistribution rights, in which case it's in
> the middle) (and, yes, Digitool is Mac-only).
>
> So, you can get demo Lisp systems on all common desktop OSs.
> You can get CMUCL on Linux, and Corman Lisp on Windows, and
> AFAIK there's nothing on Macs between the MCL demo and MCL
> itself (is Power Lisp still around?)
>
> Whether this is important depends on how you think programmers
> are going to get into Lisp. If you think they will be driven
> externally, it's no big deal. If you think that a grass-roots
> Lisp movement is necessary, it's a problem.
>
But, it's only a problem if those learning the language need a high performance system (presuming performance is the main justification
for the commercial products).
The bigger problem this newbie is having is finding an implementation that both matches the available books and can be installed
reasonably. Not everyone has Intel stuff. It was annoying to find the stuff built-in to most Linux distros is not up to the job (for
neither Common LISP nor Scheme, no less).
I find that CLISP is so far the winner.
If CLISP isn't the fastest, my current take is "so what." Once I get into a serious, heavyweight project, then I'll worry about
performance. Right now, the problem is "how do I learn this thing" and CLISP is, in my judgement, sufficient enough to lead whatever
grass roots revolution someone wants. Now, if its function is deficient, in a way not yet obvious to this newbie, that's a whole
'nother argument.
But, if the main motivation for the commercial products is performance, that isn't yet relevant to me.
Larry
Larry Loen <······@us.ibm.com> writes:
> >
> > The question is how much is reasonable to charge so that non-professionals
> > are interested.
>
> Or professionals using other languages who decide to take a look at LISP.
The usual approach here is to ask for an evalutation copy. These are
available from all the Lisp vendors I've talked to.
--tim
>>>>> "TB" == Tim Bradshaw <···@tfeb.org> writes:
>> The question is how much is reasonable to charge so that
>> non-professionals are interested.
>>
>> Or professionals using other languages who decide to take a
>> look at LISP.
TB> The usual approach here is to ask for an evalutation copy.
TB> These are available from all the Lisp vendors I've talked to.
-OR-, you can use any of several Free Software Common Lisps, and then
you own it, and then you can fix it if it breaks, and then you can
give it away to friends instead of just imploring that they try Lisp,
and then you are a hero, and then you win.
~ESP
--
Evan Prodromou
····@prodromou.san-francisco.ca.us
"James Hague" <···········@volition-inc.com> writes:
> X-no-archive: yes
>
> Evan Prodromou wrote:
> >
> > -OR-, you can use any of several Free Software Common Lisps, and then
> > you own it, and then you can fix it if it breaks, and then you can
> > give it away to friends instead of just imploring that they try Lisp,
> > and then you are a hero, and then you win.
>
> And this comes around in a complete circle again. Please point out the
> free-as-in-speech Lisps available for Windows (a popular system, regardless
> of personal opinion) that are of the caliber of LispWorks, ACL 6, or CMUCL
> (this last one has not been ported to Windows, BTW). It must generate code
> that performs within a factor of five of ACL on any popular benchmark.
>
Where did you pluck that from? Why "must" it? Can you give me one
*documented* case of any of the Lisp systems you mention being used to
build a business/application that sold for 24 million dollars? (As
CLISP is *documented* as being used.)
The point being, that CLISP can do industry level stuff, while still
being light and flexible enough. It's an *excellent* choice for
people new to Lisp.
I've tried the free offerings from ACL and Lispworks, and the Free
Lisps, and I don't see anything compellingly attractive in the
commercial offerings (even if I disregard my ideological bias).
If I need speed with CLISP, I can do the speed hungry bits in C.
I'll bet that the deciding factor in the quality of the programs you
produce is not decided by the speed of the implentation, but by the
quality of your imagination and skill.
And guess what, dude... You can't buy those.
Regards,
Peter
Peter Wood <··········@worldonline.dk> writes:
> I've tried the free offerings from ACL and Lispworks, and the Free
> Lisps, and I don't see anything compellingly attractive in the
> commercial offerings (even if I disregard my ideological bias).
>
> If I need speed with CLISP, I can do the speed hungry bits in C.
The cure sounds worse than the disease. If you're talking about
making Free Software that's portable to Windows, then I guess you have
no choice. However, if you're making commercial software, and you
can't get the speed you need from CLISP, then moving to, say,
Lispworks sounds like a fine idea. (In the context of making software
you're going to sell,) it's cheap and it's portable. Besides, if I
remember correctly, CLISP is GPL'ed and defines a "derived work" to
exclude CL programs using standard features, but to include CL
programs using CLISP-specific extensions. Wouldn't that include the
FLI?
I will agree that CLISP is a great starting place. I often start
developing with CLISP and only move to CMUCL if I need more speed or a
better FLI. In the windows world, that move would be to a commercial
lisp.
"Peter Wood" <··········@worldonline.dk> wrote in message ···················@localhost.localdomain...
> "James Hague" <···········@volition-inc.com> writes:
>
>
> I'll bet that the deciding factor in the quality of the programs you
> produce is not decided by the speed of the implentation, but by the
> quality of your imagination and skill.
>
> And guess what, dude... You can't buy those.
>
Well, not yet. They'll probably be written in Lisp, though.
But I wouldn't start holding my breath just yet... :-)
Geoff
Peter Wood wrote:
> "James Hague" <···········@volition-inc.com> writes:
>
> I've tried the free offerings from ACL and Lispworks, and the Free
> Lisps, and I don't see anything compellingly attractive in the
> commercial offerings (even if I disregard my ideological bias).
It lasts some time until you find the things that make e. g. LispWorks so
much more productive.
LispWorks...
- is better documented (compared to CLISP and CMUCL)
- includes industrial strength crossplattform GUI-Toolkits (CAPI, CLIM)
- includes crossplatform OpenGL embedded nicely in CAPI
- includes a sophisticated IDE (particularily the debugger is really nice!)
- Has a better FLI than CMUCL or CLISP
- Has better MOP-support
- Can deliver executables
I love to simply click on a debugger-stackframe and get my Editor-Window
with the corresponding expression highlighted.
I love those many little graphical browsers and tools - yes you have to get
used to how LispWorks works but then it is IMHO far more productive than
the free Lisps.
> If I need speed with CLISP, I can do the speed hungry bits in C.
This would be easier in the commercial Lisps too...
Regards,
Jochen Schmidt
* Peter Wood wrote:
> Where did you pluck that from? Why "must" it? Can you give me one
> *documented* case of any of the Lisp systems you mention being used to
> build a business/application that sold for 24 million dollars? (As
> CLISP is *documented* as being used.)
I could, but I would have to kill you. Really, I know of such systems,
and I've worked on one, but I'm not at liberty to tell you what it is
I'm afraid.
I guess this fails on the `documented' front, but almost all
high-value software like this is covered by NDAs.
--tim
Tim Bradshaw <···@cley.com> writes:
> * Peter Wood wrote:
>
> > Where did you pluck that from? Why "must" it? Can you give me one
> > *documented* case of any of the Lisp systems you mention being used to
> > build a business/application that sold for 24 million dollars? (As
> > CLISP is *documented* as being used.)
>
> I could, but I would have to kill you. Really, I know of such systems,
> and I've worked on one, but I'm not at liberty to tell you what it is
> I'm afraid.
>
> I guess this fails on the `documented' front, but almost all
> high-value software like this is covered by NDAs.
OK, I believe you, out of the goodness of my heart. ;-)
The fact remains that CLISP was used as the implentation in a small,
start-from-scratch, two-guys-in-an-attic, business which made a lot of
money. The post I was objecting to suggested speed as the only
criterion for chosing an implementation. Yes - for most things CLISP
is slower than the others. But that doesn't always matter.
Regards,
Peter
Peter Wood <··········@worldonline.dk> writes:
> The fact remains that CLISP was used as the implentation in a small,
> start-from-scratch, two-guys-in-an-attic, business which made a lot of
> money. The post I was objecting to suggested speed as the only
> criterion for chosing an implementation. Yes - for most things CLISP
> is slower than the others. But that doesn't always matter.
I think this is a good point to make.
Speed of deployed implementation is rarely what stands between anyone and
a successful business. At the point that it is, someone will be standing
next to you with a checkbook ready to write a check for a better
implementation.
As long as the implementation conforms to the standard, it can be "second
sourced". Keep your platform dependencies isolated to a small number of
files in case you have to change platforms, and then just stop worrying
about things like speed. Choose the one that makes you feel like you can
get work done.
* Peter Wood wrote:
> The fact remains that CLISP was used as the implentation in a small,
> start-from-scratch, two-guys-in-an-attic, business which made a lot of
> money. The post I was objecting to suggested speed as the only
> criterion for chosing an implementation. Yes - for most things CLISP
> is slower than the others. But that doesn't always matter.
Oh I definitely wasn't trying to imply that speed matters. I don't
know of a CL system that is slow enough that it's likely to hurt most
applications: machines are really fast now, and a huge amount of the
things people misdiagnose as slowness are either algorithmic crapness
or enormously unbalanced configurations (the canonical 1GHz processor
and 64Mb of memory PC for instance, instead of something with a
reasonable amount of memory, like 20x as much...). Things like
standards-conformance and support are a much bigger deal to me.
--tim
On 09 Jun 2001 09:16:25 +0200, Peter Wood <··········@worldonline.dk>
wrote:
> The fact remains that CLISP was used as the implentation in a small,
> start-from-scratch, two-guys-in-an-attic, business which made a lot of
> money. The post I was objecting to suggested speed as the only
Are you referring to ViaWeb/Yahoo! Store?
Paolo
--
EncyCMUCLopedia * Extensive collection of CMU Common Lisp documentation
http://cvs2.cons.org:8000/cmucl/doc/EncyCMUCLopedia/
Peter Wood <··········@worldonline.dk> writes:
> Paolo Amoroso <·······@mclink.it> writes:
>
> > Are you referring to ViaWeb/Yahoo! Store?
>
> Yes.
Well, as embarrassing as it is to say, it was precisely speed that
caused us to lose out to Clisp in that particular situation. We
had implemented both pretty and non-pretty printing with the same
mechanism, and had also been working with Gray streams for quite a
while, so our i/o was actually slower for throughput than Clisp.
It was this failure that sparked us to do massive rewrites of our
i/o system, including a new, fast object-oriented streams system
and non-pretty printing that didn't force the use of pretty-streams
and dispatch tables. Result: 10x to 40x speedup in non-pretty
printing situations, and 2x to 4x in pretty-printing situations.
The point is that indeed speed was critical in selecting the
implementation (it just didn't go our way that time).
--
Duane Rettig Franz Inc. http://www.franz.com/ (www)
1995 University Ave Suite 275 Berkeley, CA 94704
Phone: (510) 548-3600; FAX: (510) 548-8253 ·····@Franz.COM (internet)
Peter Wood <··········@worldonline.dk> wrote:
: The fact remains that CLISP was used as the implentation in a small,
: start-from-scratch, two-guys-in-an-attic, business which made a lot of
: money. The post I was objecting to suggested speed as the only
: criterion for chosing an implementation. Yes - for most things CLISP
: is slower than the others. But that doesn't always matter.
It matters, because the Yahoo guys who took over the store editor were
forced to rewrite it. The bad point is that they did in C from scratch
instead of simply porting it to a faster lisp. Of course this is only their
private problem, but I'm afraid of rumors again (as it also happened with
Garnet et al.)
--
Reini Urban
http://xarch.tu-graz.ac.at/acadwiki/AutoLispFaq
>>>>> "James" == James Hague <···········@volition-inc.com> writes:
James> And this comes around in a complete circle again. Please point
James> out the free-as-in-speech Lisps available for Windows (a
James> popular system, regardless of personal opinion)
I haven't got any benchmarks nor do I know if it would work out of the
box on Windows, but there are at least two systems based on CL->C
compilation, and there should be no major problems in making that
work on Windows.
I am here thinking about GCL:
ftp://rene.ma.utexas.edu/pub/gcl
and ECL, which has been brought back to life in Spain, I think:
http://www.arrakis.es/~worm/ecls.html
(better references probably exists, but this was what I could dig up
in a hurry.)
------------------------+-----------------------------------------------------
Christian Lynbech | Ericsson Telebit, Skanderborgvej 232, DK-8260 Viby J
Phone: +45 8938 5244 | email: ·················@ted.ericsson.dk
Fax: +45 8938 5101 | web: www.ericsson.com
------------------------+-----------------------------------------------------
Hit the philistines three times over the head with the Elisp reference manual.
- ·······@hal.com (Michael A. Petonic)
>>>>> "JH" == James Hague <···········@volition-inc.com> writes:
JH> And this comes around in a complete circle again. Please
JH> point out the free-as-in-speech Lisps available for Windows (a
JH> popular system, regardless of personal opinion)
Sure, but it's not Free. I don't particularly care about Windows.
I have 3 free CLs on my Debian box: CLISP, CMUCL, and gcl. Two make
native code. This seems pretty OK to me.
I'd bet, but I don't know, that gcl would build on cygwin. Why don't
you give it a try?
~ESP
--
Evan Prodromou
····@prodromou.san-francisco.ca.us
>>>>> "David" == David Thornley <········@visi.com> writes:
David> (and, yes, Digitool is Mac-only).
Not any more. They have released the source code of the runtime
environment as Open Source and there is a port of it to LinuxPPC
(it works, I have tried it).
Check out
http://www.clozure.com/openmcl
for more information.
MCL is of course still PPC-only, I would guess that porting the
compiler would be a major undertaking.
------------------------+-----------------------------------------------------
Christian Lynbech | Ericsson Telebit, Skanderborgvej 232, DK-8260 Viby J
Phone: +45 8938 5244 | email: ·················@ted.ericsson.dk
Fax: +45 8938 5101 | web: www.ericsson.com
------------------------+-----------------------------------------------------
Hit the philistines three times over the head with the Elisp reference manual.
- ·······@hal.com (Michael A. Petonic)
Christian Lynbech wrote:
>
> MCL is of course still PPC-only, I would guess that porting the
> compiler would be a major undertaking.
As I remeber there was a port to SPARC!
Jochen Schmidt wrote:
> As I remeber there was a port to SPARC!
From the Opensource MCL FAQ:
At one point - because of concerns about portability to other
development platforms - MCL was ported to Solaris/Sparc. (It worked
well enough to compile itself and to cross-compile for the PPC; once
the proof-of-concept had been proven, the Sparc port wasn't maintained.)
I suppose porting it is doable, but currently PPC seems to be the only platform.
//Lennart
"James Hague" <···········@volition-inc.com> writes:
> Janis Dzerins wrote:
> >
> > Xanalys LispWorks Professional with IDE is just a little bit above in
> > price and *very* much above in all other aspects. And it also includes
> > CLIM.
> >
> > And free Clisp or CMUCL hold their own no worse than gcc.
>
> There is too much explaining away of this, methinks. Let's see some
> benchmarks comparing CMUCL, Clisp, LispWorks, and ACL. I have been under
> the impression that (A) CMUCL is not available for Windows, and (B) Clisp is
> at least an order of magnitude slower than the slowest of the machine code
> generating Lisps available for Windows.
Can you use gcc on windows then, to produce native windows code?
Note that there are free Lisp systems from Xanalys, Franz, and
Roger Corman (if you can live with the restrictions of these packages).
--
Raymond Wiker
·············@fast.no
From: James Hague
Subject: Re: a bit of history of lisp
Date:
Message-ID: <3b16a265$1_4@newsfeeds>
Raymond Wiker wrote:
>
> Note that there are free Lisp systems from Xanalys, Franz, and
> Roger Corman (if you can live with the restrictions of these packages).
Nit: Franz stopped having a free version of ACL a while back. You can get
a 30 day license, but that's it.
James
-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----== Over 80,000 Newsgroups - 16 Different Servers! =-----
"James Hague" <···········@volition-inc.com> writes:
> Raymond Wiker wrote:
> >
> > Note that there are free Lisp systems from Xanalys, Franz, and
> > Roger Corman (if you can live with the restrictions of these packages).
>
> Nit: Franz stopped having a free version of ACL a while back. You can get
> a 30 day license, but that's it.
>
> James
Nit nit: You can get such licenses for six months.
--
Duane Rettig Franz Inc. http://www.franz.com/ (www)
1995 University Ave Suite 275 Berkeley, CA 94704
Phone: (510) 548-3600; FAX: (510) 548-8253 ·····@Franz.COM (internet)
* Duane Rettig <·····@franz.com>
| > Nit: Franz stopped having a free version of ACL a while back. You can get
| > a 30 day license, but that's it.
| >
| > James
|
| Nit nit: You can get such licenses for six months.
And then you can email them and get another six months. At least I
did.
--
Vebjorn Ljosa
>>>>> "Raymond" == Raymond Wiker <·············@fast.no> writes:
Raymond> "James Hague" <···········@volition-inc.com> writes:
>> Janis Dzerins wrote:
>> >
>> > Xanalys LispWorks Professional with IDE is just a little bit above in
>> > price and *very* much above in all other aspects. And it also includes
>> > CLIM.
>> >
>> > And free Clisp or CMUCL hold their own no worse than gcc.
>>
>> There is too much explaining away of this, methinks. Let's see some
>> benchmarks comparing CMUCL, Clisp, LispWorks, and ACL. I have been under
>> the impression that (A) CMUCL is not available for Windows, and (B) Clisp is
>> at least an order of magnitude slower than the slowest of the machine code
>> generating Lisps available for Windows.
Raymond> Can you use gcc on windows then, to produce native windows code?
Yes. It's called djgcc (I think) or you can get Cygwin which has gcc.
Ray
Raymond Toy <···@rtp.ericsson.se> writes:
> >>>>> "Raymond" == Raymond Wiker <·············@fast.no> writes:
>
> Raymond> "James Hague" <···········@volition-inc.com> writes:
> >> Janis Dzerins wrote:
> >> >
> >> > Xanalys LispWorks Professional with IDE is just a little bit above in
> >> > price and *very* much above in all other aspects. And it also includes
> >> > CLIM.
> >> >
> >> > And free Clisp or CMUCL hold their own no worse than gcc.
> >>
> >> There is too much explaining away of this, methinks. Let's see some
> >> benchmarks comparing CMUCL, Clisp, LispWorks, and ACL. I have been under
> >> the impression that (A) CMUCL is not available for Windows, and (B) Clisp is
> >> at least an order of magnitude slower than the slowest of the machine code
> >> generating Lisps available for Windows.
>
> Raymond> Can you use gcc on windows then, to produce native windows code?
>
> Yes. It's called djgcc (I think) or you can get Cygwin which has gcc.
Actually, djgcc is gcc for DOS. gcc on windows comes with a UNIX
emulation layer (Cygwin) or without (Mingwin).
··@panix.com (Paul Wallich) writes:
> I'm not sure the above analyses are correct. What many dotcoms
> produced in terms of web sites (or tools for them) was not
> significantly more useful than your typical prototype AI
> project. The difference was more that Web stuff was radiating into
> an essentially new space, whereas the AI folks had to compete with
> existing applications folks who had 20-30 years of really good
> experience in pushing back. (In some ways, both foundered not so
> much on inability to deliver useful stuff as inability to find
> people willing to _pay_ for things they could in fact deliver.)
I think I disagree with this, but it may be a matter of opinion. I
came across too many AI people who claimed that they could solve -
given just a little funding - completely open problems like natural
language translation or something, and I never came across dot-com
people who claimed that - their claims always seemed just economically
implausible.
> In addition, the AI bubble was never nearly as big in dollar terms
> as it was in hype. Tens of millions, maybe hundreds of millions at
> the outside (I don't think DARPA et al ever got over about $100M a
> year).
I definitely agree with this (in fact this was my original point) -
the recent/ongoing collapse is *much* bigger than almost anything
financially.
--tim
In article <···············@tfeb.org>, Tim Bradshaw <···@tfeb.org> wrote:
>··@panix.com (Paul Wallich) writes:
>
>> I'm not sure the above analyses are correct. What many dotcoms
>> produced in terms of web sites (or tools for them) was not
>> significantly more useful than your typical prototype AI
>> project. The difference was more that Web stuff was radiating into
>> an essentially new space, whereas the AI folks had to compete with
>> existing applications folks who had 20-30 years of really good
>> experience in pushing back. (In some ways, both foundered not so
>> much on inability to deliver useful stuff as inability to find
>> people willing to _pay_ for things they could in fact deliver.)
>
>I think I disagree with this, but it may be a matter of opinion. I
>came across too many AI people who claimed that they could solve -
>given just a little funding - completely open problems like natural
>language translation or something, and I never came across dot-com
>people who claimed that - their claims always seemed just economically
>implausible.
Different research areas :-)
The AI people I talked to were pretty clear on the things they could
and couldn't solve, albeit there sometimes was some naivete about
what proportion of the real world the solvable problems constituted...
(Sometimes/often constraining the set of possible inputs is just fine
-- lots of naive funders, alas, were asking for superhuman rather than
human-level performance, and asking for it in the absence of interactivity;
i.e. the NLP software wasn't allowed to say, "Huh? could you repeat that?")
>> In addition, the AI bubble was never nearly as big in dollar terms
>> as it was in hype. Tens of millions, maybe hundreds of millions at
>> the outside (I don't think DARPA et al ever got over about $100M a
>> year).
>
>I definitely agree with this (in fact this was my original point) -
>the recent/ongoing collapse is *much* bigger than almost anything
>financially.
A consequence both of economic growth and Moore's Law. (One of
the really interesting things is how many Really Stupid Algorithms
for, say, translation have become practical because you can simply
store all common tourist utterances on a $250 disk that might well
hold more data than the RAM of all the 80's lisp machines in the
world combined.)
paul
In article <···············@tfeb.org>, Tim Bradshaw <···@tfeb.org> wrote:
>Duane Rettig <·····@franz.com> writes:
>
>> Michael Parker <·······@pdq.net> writes:
>
>> > That would have been the "AI Winter". AI and Lisp shops were in
>> > roughly the roughly same position as the dot-com companies of more
>> > recent note, and fell from grace similarly. The markets and venture
>> > capitalists abandoned them when AI failed to live up to their
>> > unrealistic expectations.
>>
>> It was worse than that; many large companies sunk billions into AI,
>
>I dunno if it was worse in any reasonable sense: a huge amount of
>money has been wiped off stockmarkets recently, and probably a fair
>amount of that can be put down to ludicrous expectations of new
>technology leading to completely unrealistic valuations of companies -
>not just small dot-coms, either.
One significant difference, IMHO, was that Lisp and AI were much more
expensive technologies. If you want to do e-commerce now, you purchase a
few servers, hire a webmaster, some designers (there are zillions of them
out there looking for jobs), and a database guy, and off you go; most
companies outsource most of this, and the cost is pretty reasonable.
But Lisp and AI in the 80's were pretty expensive. To get decent
performance and development environments you had to buy expensive Lisp
Machines (minimal configurations cost $50-100K); Lisp on general purpose
workstations didn't catch up until the early 90's, and even then you had to
use a $10-50K Sun, SGI, or HP system rather than a cheap PC. And Lisp
programmers were never in plentiful supply, even during the AI heyday,
since AI was generally in the graduate-level curriculum (at MIT we had one
undergraduate Introduction to AI class), so staffing was also quite
expensive.
--
Barry Margolin, ······@genuity.net
Genuity, Burlington, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Please DON'T copy followups to me -- I'll assume it wasn't posted to the group.
Barry Margolin <······@genuity.net> writes:
>
> One significant difference, IMHO, was that Lisp and AI were much more
> expensive technologies. If you want to do e-commerce now, you purchase a
> few servers, hire a webmaster, some designers (there are zillions of them
> out there looking for jobs), and a database guy, and off you go; most
> companies outsource most of this, and the cost is pretty reasonable.
>
I think that's right, even though it's kind of obviously wrong - you
could argue that you actually need to buy an E10k and an Oracle
license for it, which is huge money - much more than a lisp machine.
But I think it's right because the entry cost is low even if scaling
kills you later (and of course many/most dot-coms died before they
scaled), and you can also get pretty far with smaller kit anyway.
Outsourcing also lets you make the costs look small, though I suspect
that's often because people are not costing things right (akin to the
`Windows PCs are cheap ... if you ignore the 10% of their time that
everyone in the company spends ineptly managing their PC' fallacy)
--tim
Duane Rettig wrote:
>
> Not all lisp-based shops crashed or burned; some of them went underground.
> Some of them still are, and don't mention that they get their job done
> quickly _because_ there is lisp inside. In fact, some keep this fact
> quiet as a competitive advantage.
>
This is a persistent rumor in the Lisp community. Many of us hope it's
true (I do!). I only know of one documented case, though - Graham's
Viaweb/Yahoo Store. Does anyone know of others that can be safely,
ethically documented? I know about the success stories on vendor web
sites - these are mostly about applications that were delivered in Lisp
"above ground", so they don't fit the underground band of hackers image.
Please note, I am *NOT* asking people to reveal still-valuable business
secrets. DON'T post unless the information is out in the public
elsewhere, or you are one of the people doing this and want to uncloak.
>
> I for one am forthcoming because I believe the worst is past us, and
> lisp is once again on the rise, slowly and steadily.
>
I dearly hope you're right. Now, if we can get the rest of the world to
punt XML in favor of S-expressions...
--
Remove obvious stuff to e-mail me.
Bob Bane
On Tue, 29 May 2001 13:00:57 -0400, Bob Bane <····@removeme.gst.com> wrote:
>Duane Rettig wrote:
>>
>> Not all lisp-based shops crashed or burned; some of them went underground.
>> Some of them still are, and don't mention that they get their job done
>> quickly _because_ there is lisp inside. In fact, some keep this fact
>> quiet as a competitive advantage.
>>
>
>This is a persistent rumor in the Lisp community. Many of us hope it's
>true (I do!). I only know of one documented case, though - Graham's
>Viaweb/Yahoo Store. Does anyone know of others that can be safely,
>ethically documented?
I wrote a clinical decision support system which is used in hospitals around the
US and several other countries (such as Memorial Sloan Kettering in NY) and is
considered the leading system of its kind. It is part of the Eclipsys Sunrise
Clinical Manager (formerly HealthVISION's CareVISION product) which is a medical
records system. This system is discussed by Bill Gates in Business @ the Speed
of Thought (I know, big whoopee... :-)
The clinical decision support allows rules to be written in a language called
Arden (an international standard for medical logic) and includes an Arden
compiler and a rules engine. I implemented this compiler to translate all the
rules to Lisp at run time, and run them in a lisp interpreter (which I wrote)
which is embedded in the system. It works very well, is easy to debug (you can
get a lisp console--undocumented), includes an FFI for calling C functions, an
ODBC interface for database queries etc. Using Lisp allowed me to get the first
version up and running in 12 weeks from starting. It helped in the long
run--never any problems from memory leaks, and performance was always excellent.
I didn't exactly keep use of lisp a secret, but didn't talk about it much
either. Typically we would tell customers that it was in C++ (the lisp
interpreter was in C++). Never bothered to generate native code because 95% of
the time was typically spent in the database queries.
I really believed the use of Lisp allowed us to leapfrog ahead of our
competitors, some of whom had spent years designing such a product, with many
more engineers (and we beat them to market with two of us working for 6 months).
Unfortunately I didn't have enough stock to benefit much financially from this,
but it was satisfying.
Roger
>>>>> "bob" == Bob Bane <····@removeme.gst.com> writes:
bob> Please note, I am *NOT* asking people to reveal still-valuable
bob> business secrets. DON'T post unless the information is out in
bob> the public elsewhere, or you are one of the people doing this
bob> and want to uncloak.
I wouldn't otherwise mention it, but I have already read on this
newsgroup several times that Gensym uses Chestnut, which is a
Lisp-to-C translator that generates quite readable C code from Common
Lisp.
If you think about it, if this is sufficient for a company to use,
then you can get a Lisp implementation (commercial, free, etc.),
develop with it, but then, at the end of the day, translate it to C
and use gcc (or whatever C compiler is best for you), and pay *no*
royalties, and can get very efficient code. It's easy to call into
and out of, just like C is in general, and (if it works) you save
*tons* on licensing if you sell at high volumes. You just pay for
development licenses, where what you're paying for is a nice
development environment, debugger, inspector, maybe an editor,
profiler (though that would not correlate perfectly with the
translated code), etc.
dave
On Mon, 28 May 2001 21:18:48 -0500, Michael Parker <·······@pdq.net> wrote:
> That would have been the "AI Winter". AI and Lisp shops were in
> roughly the roughly same position as the dot-com companies of more
> recent note, and fell from grace similarly. The markets and venture
Incidentally, nobody seems to be blaming technologies such as HTML or HTTP
for the recent dot-com shake-out. But after the AI Winter, Lisp was blamed
for the failures of the AI industry.
Paolo
--
EncyCMUCLopedia * Extensive collection of CMU Common Lisp documentation
http://cvs2.cons.org:8000/cmucl/doc/EncyCMUCLopedia/
>>>>> "Paolo" == Paolo Amoroso <·······@mclink.it> writes:
Paolo> On Mon, 28 May 2001 21:18:48 -0500, Michael Parker <·······@pdq.net> wrote:
>> That would have been the "AI Winter". AI and Lisp shops were in
>> roughly the roughly same position as the dot-com companies of more
>> recent note, and fell from grace similarly. The markets and venture
Paolo> Incidentally, nobody seems to be blaming technologies such as HTML or HTTP
Paolo> for the recent dot-com shake-out. But after the AI Winter, Lisp was blamed
Paolo> for the failures of the AI industry.
Were there any major Lisp successes before and/or during the AI
Winter? HTML and HTTP were pretty successful with lots of hype before
all the dot-coms got in to it.
Ray
Raymond Toy <···@rtp.ericsson.se> writes:
> Were there any major Lisp successes before and/or during the AI
> Winter? HTML and HTTP were pretty successful with lots of hype before
> all the dot-coms got in to it.
Because of the large investment, I think it usually happened in-house
rather than in private ventures, so it's harder to see. It's not like
companies now, many of which are one-to-one with the products they made.
You're talking now about the days when dinosaurs (like IBM) still ruled
the earth.
I worked for Schlumberger in 1979 on a project with Dave Barstow from
Stanford where we used Lisp/AI technology involving a scenario where
you drop a thing down a potential oil well, bombard the surrounding
soil with neutrons, get off gamma ray spectra, and make a determination
about whether you can get oil from the place. Experts were few, and
hard to clone in human form, so we created artificial experts.
I didn't follow the project in detail later, though I think the resulting
item--the DipMeter Advisor or some such thing--was successful enough
that Schlumberger later founded its Palo Alto research center.
I think similar projects sprung up all over the place, and two
products in particular, KEE and ART, were expert systems shells sold
by companies that thrived for at least a while---Intellecorp and
Inference, if I recall correctly. I think ART is still available in
some form today, though ported to other languages, probably partly due
to AI Winter.
(One of the paradoxes of Lisp is that although it has traditionally
enabled the production of products that it is believed by many could
not have been done in other languages, once those projects are
created, the product is often possible to port to other languages.
Lisp allows important flexibility in design and development that is
not always needed at runtime. This is, I think, not accidentally
related to the often-noted aspect of AI that something always seems
like AI until you know how to do it--then it's just programming.)
I think Macsyma, written in Lisp, was commercially successful. I was
told by someone that it was the only product Symbolics sold that was a
consistent money-maker, paying for its own staff, throughout its
product lifetime. This was, I think, why it managed to later spin off
as its own company and survive for a while there, too.
And there were a lot of uses of Lisp in managing planning and
scheduling while I was at Symbolics. I think a number of airline
companies did and some still do use Lisp for planning and scheduling
flights and/or maintenance. Power companies, too. But these again were
not companies selling Lisp that went public and made Lisp a big name,
they were the integration of Lisp into a big company's core technology
in a way that the public mostly didn't see or appreciate. And I'm sure
at least some of them ditched this perfectly fine technology as soon as
AI winter hit, rather than defend its ongoing use.
It's even sadder than that, I'm afraid...
A medium-sized company sucessfully used the Concept Modeller (a
Lisp-based expert system shell similar to ICAD) to implement a automated
design /accurate bid estimator. This company is still (as of as
recently as a year ago, last I spoke to them) using this system
profitably to generate accurate bids from the automated design system
which can are then used to generate a final design. The system runs on
old Sun Sparcs.
And now the sad part. Upper management has no clue of the value of this
system, having spent virtually no money on it in 10 years. No one ever
ran the numbers on the engineering manhours saved or the amount of money
made by avoiding bad bids (both in terms of exposure for too-low bids
and lost sales for too-high bids). I suspect the maintainance plan is
to shop around flea markets for old Sparcs.
Kent M Pitman wrote:
>
<snip>
> I think similar projects sprung up all over the place, and two
> products in particular, KEE and ART, were expert systems shells sold
> by companies that thrived for at least a while---Intellecorp and
> Inference, if I recall correctly. I think ART is still available in
> some form today, though ported to other languages, probably partly due
> to AI Winter.
<snip>
> in a way that the public mostly didn't see or appreciate. And I'm sure
> at least some of them ditched this perfectly fine technology as soon as
> AI winter hit, rather than defend its ongoing use.
On Tue, 29 May 2001 16:24:54 -0700, Raymond Laning
<········@west.raytheon.com> wrote:
>It's even sadder than that, I'm afraid...
>
>A medium-sized company sucessfully used the Concept Modeller (a
>Lisp-based expert system shell similar to ICAD) to implement a automated
>design /accurate bid estimator. This company is still (as of as
>recently as a year ago, last I spoke to them) using this system
>profitably to generate accurate bids from the automated design system
>which can are then used to generate a final design. The system runs on
>old Sun Sparcs.
(Btw, your host address made me remember the first computer I really
dirtied my hands on: A Raytheon 704, 1 Mhz , 8 kb ferrite core memory,
no disk, only a teletype and paper tape reader ...)
We are getting near my application domains at that time, ie
mechanical design, production planning, etc. These domains are
different from Kent's examples (finding oil, planning and scheduling
for airlines) or Roger's (clinical decision support), which have a
fairly stable and generally applicable knowledge base to mine.
But many "AI" areas depend on getting knowledge - rules, data, what
have you - locally from people in your organization, and the cost of
that task was completely under-estimated (if it could be done at all),
and the knowledge changes more rapidly.
Another main obstacle was changing the methods and the way of thinking
used by designers, planners, etc. To use a current buzzword: their
competence profile was unsuitable.
>And now the sad part. Upper management has no clue of the value of this
>system, having spent virtually no money on it in 10 years. No one ever
>ran the numbers on the engineering manhours saved or the amount of money
>made by avoiding bad bids (both in terms of exposure for too-low bids
>and lost sales for too-high bids). I suspect the maintainance plan is
>to shop around flea markets for old Sparcs.
Yet another difficulty was to demonstrate the overall profitability
when using these techniques. The total cost is of course _much_ higher
than hardware and program development costs. Many of us, working as
technical experts and programmers, failed to recognize this.
And still do, I think. We put excellent software on the counter, only
to find that there are few buyers. That is not surprising if one
remembers that beauty occurs in the eye of the beholder. :)
/Lars
On Tue, 29 May 2001, Kent M Pitman wrote:
> I think Macsyma, written in Lisp, was commercially successful. I was
> told by someone that it was the only product Symbolics sold that was a
> consistent money-maker, paying for its own staff, throughout its
> product lifetime. This was, I think, why it managed to later spin off
> as its own company and survive for a while there, too.
>
> And there were a lot of uses of Lisp in managing planning and
> scheduling while I was at Symbolics. I think a number of airline
> companies did and some still do use Lisp for planning and scheduling
> flights and/or maintenance. Power companies, too. But these again were
> not companies selling Lisp that went public and made Lisp a big name,
> they were the integration of Lisp into a big company's core technology
> in a way that the public mostly didn't see or appreciate. And I'm sure
> at least some of them ditched this perfectly fine technology as soon as
> AI winter hit, rather than defend its ongoing use.
In this vein, let's not forget Symbolics' S-World 3D graphics and
rendering software. This later became Nichimen's N-World and Mirai
products, now maintained by Winged Edge. I don't know details of its
profitability, but it was used for major video games (like Nintendo's
Super Mario 64) and has been around an awfully long time. Nichimen/Winged
Edge doesn't make any secret of it being written in Lisp; in fact, easy
debuggability and customizability is touted as an advantage.
Tim
In article <····························@4ax.com>,
Paolo Amoroso <·······@mclink.it> wrote:
>Incidentally, nobody seems to be blaming technologies such as HTML or HTTP
>for the recent dot-com shake-out. But after the AI Winter, Lisp was blamed
>for the failures of the AI industry.
See my post regarding the expense of doing Lisp in the 80's. I think that
explains it.
--
Barry Margolin, ······@genuity.net
Genuity, Burlington, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Please DON'T copy followups to me -- I'll assume it wasn't posted to the group.
Barry Margolin <······@genuity.net> writes:
> See my post regarding the expense of doing Lisp in the 80's. I think that
> explains it.
Doing java in the late nineties wasn't exactly inexpensive either: Armies
of over-payed and under-qualified consultants putting together e-something
java applications that were so slow that they - in order to support even
a ridiculously low number of customers - required Sun hardware that
was more expensive than a Xerox Lisp Machine used to be in the late
eighties...
--
(espen)
>>>>> "Espen" == Espen Vestre <·····@*do-not-spam-me*.vestre.net> writes:
Espen> Barry Margolin <······@genuity.net> writes:
>> See my post regarding the expense of doing Lisp in the 80's. I think that
>> explains it.
Espen> Doing java in the late nineties wasn't exactly inexpensive
Espen> either...
And while ordinary Java was free from Sun, other variations weren't.
When I joined my current employer in '98 we were just starting up a
project to provide us (then an independent router vendor, bought last
year by Ericsson) with a new network management framework for our
routers.
It was initially decided to do it in Java, but (forunately enough, he he he)
Sun demanded ridiculeus amounts of money for their Embedded Java Edition.
We went with Guile Scheme instead, as proposed by a newly employed
lisp nut (me).
------------------------+-----------------------------------------------------
Christian Lynbech | Ericsson Telebit, Skanderborgvej 232, DK-8260 Viby J
Phone: +45 8938 5244 | email: ·················@ted.ericsson.dk
Fax: +45 8938 5101 | web: www.ericsson.com
------------------------+-----------------------------------------------------
Hit the philistines three times over the head with the Elisp reference manual.
- ·······@hal.com (Michael A. Petonic)
Paolo Amoroso <·······@mclink.it> writes:
> On Mon, 28 May 2001 21:18:48 -0500, Michael Parker <·······@pdq.net> wrote:
>
> > That would have been the "AI Winter". AI and Lisp shops were in
> > roughly the roughly same position as the dot-com companies of more
> > recent note, and fell from grace similarly. The markets and venture
>
> Incidentally, nobody seems to be blaming technologies such as HTML or HTTP
> for the recent dot-com shake-out. But after the AI Winter, Lisp was blamed
> for the failures of the AI industry.
That's because
(a) everyone knows html works
(b) no one knows what http is
(c) everyone knows the four letters ".com" caused it
All they need is one scapegoat. Blaming ".com" is the moral
equivalent of blaming lisp. They've pointed to an inanimate object
that has hexed their ability to do good management.
(Pity the person who owns lisp.com !)
Surely nothing to do with the "new economics" caused the recent
downfall, for example. People wandering around saying that companies
don't have to turn a profit to be valuable in this new era could not
have had anything to do with the crash.
This is one reason that I'm skeptical of free software as a business
model (discussed amply in other threads, please no followup in this
thread except as to my specific point about scapegoating). Free
software, too, purports to rearrange traditional business models and
yet still win, teling you that if you spend money on something and
then make that thing freely available to your competitors, you will
remain as strong as your competitors, who may spend money on things
you don't have access to. I claim the math does not balance, and I
don't think I'm alone in that. But my specific reason for mentioning
it here is not to push that point redundantly, but rather to note that
I fully expect that as failures happen in that market which might be
attributed to this bad business model, people will try to deflect
attention from what might be the actual cause. This will both avoid
people pointing to the individuals who were responsible for the
underlying bad business decision, AND will make it hard to collect
statistics about the real reason for failure.
Kent M Pitman <······@world.std.com> writes:
> All they need is one scapegoat. Blaming ".com" is the moral
> equivalent of blaming lisp. They've pointed to an inanimate object
> that has hexed their ability to do good management.
Wouldn't it have been a nice (and cruel) revenge if ".com companies"
had been "java companies" (or perl-, C++-, VBS-, whatever) instead?
> Surely nothing to do with the "new economics" caused the recent
> downfall, for example. People wandering around saying that companies
> don't have to turn a profit to be valuable in this new era could not
> have had anything to do with the crash.
Sigh, it's almost unbelievable that it's just a little more than
a year ago that lots of pointy-haired bosses even in fairly large
companies would say and believe those things. And now, of course,
it's the complete opposite: companies are performing the most
ridiculously conservative actions just to look old-fashioned enough
for the stock market :-(
--
(espen)
[I'm mainly commenting on Andrei's claim quoted by Robert - Paolo]
On Mon, 28 May 2001 15:45:13 GMT, ······@nextra.sk (Robert Vazan) wrote:
> I heard somebody (Andrei Alexandrescu) somewhere
> (comp.lang.c++.moderated) saying something (precise quote at
> end of post) about major crash of lisp-based shops in 80's.
[...]
> ****** begin quote (Andrei Alexandrescu) *******
>
> "Thant Tessman" <·····@acm.org> wrote
> > Matt Seitz wrote:
> > > [...] Otherwise, at best it merely serves to convince current
> > > Lisp programmers to not try C++.
> >
> > Believe me, they don't need any convincing.
>
> I guess all guys who got laid off after the major crash and
> burn of all
> LISP-based shops in the eighties didn't, either. Besides, of
Major crash and burn of all Lisp-based shops? If by "shops" Andrei refers
to Lisp tools vendors, some major early Lisp shops are actually still in
business:
Franz
http://www.franz.com/
Gold Hill
http://www.goldhill-inc.com/
Xanalys (formerly Harlequin)
http://www.xanalys.com
And some well-known early Lisp products by companies that no longer exist
are still commercially available from other vendors:
Lucid Common Lisp (originally by Lucid, now sold as Liquid Common Lisp by
Xanalys)
http://www.xanalys.com
Coral Common Lisp (originally by Coral, now sold as Macintosh Common Lisp
by Digitool - please correct this information if it's not accurate)
http://www.digitool.com
Paolo
--
EncyCMUCLopedia * Extensive collection of CMU Common Lisp documentation
http://cvs2.cons.org:8000/cmucl/doc/EncyCMUCLopedia/
In article <····························@4ax.com>,
Paolo Amoroso <·······@mclink.it> wrote:
>Major crash and burn of all Lisp-based shops? If by "shops" Andrei refers
>to Lisp tools vendors, some major early Lisp shops are actually still in
>business:
And many dot-coms are still in business, too. But that doesn't contradict
the fact that the industry crashed a year ago. If an industry is decimated
(the term comes from a root meaning "one-tenth", i.e. reduced to 1/10th its
size), that's a major crash and burn.
And the companies that weren't totally wiped out were hurt badly. They're
not as big, or they had to branch out into other businesses.
In the case of Harlequin (now Xanalys), Lisp wasn't even their main product
to begin with.
--
Barry Margolin, ······@genuity.net
Genuity, Burlington, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Please DON'T copy followups to me -- I'll assume it wasn't posted to the group.
Barry Margolin <······@genuity.net> writes:
>
> And many dot-coms are still in business, too. But that doesn't contradict
> the fact that the industry crashed a year ago. If an industry is decimated
> (the term comes from a root meaning "one-tenth", i.e. reduced to 1/10th its
> size), that's a major crash and burn.
>
`Decimate' actually originally meant to reduce *by* one tenth, though
I think its meanign is changing.
--tim
Tim wrote:
> `Decimate' actually originally meant to reduce *by* one tenth, though
> I think its meanign is changing.
As in to kill one in ten -- a Roman punishment for Army mutineers.
:)w
I believe it referred to a roman legionary discipline in which a members of
a military troupe were executed, 1 man in 10, as a punishment for a grave
infraction
Jeff
Tim Bradshaw <···@tfeb.org> wrote in message news:<···············@tfeb.org>...
> Barry Margolin <······@genuity.net> writes:
>
> >
> > And many dot-coms are still in business, too. But that doesn't contradict
> > the fact that the industry crashed a year ago. If an industry is decimated
> > (the term comes from a root meaning "one-tenth", i.e. reduced to 1/10th its
> > size), that's a major crash and burn.
> >
>
> `Decimate' actually originally meant to reduce *by* one tenth, though
> I think its meanign is changing.
>
> --tim
On Tue, 29 May 2001 15:54:12 GMT, Barry Margolin <······@genuity.net>
wrote:
> In article <····························@4ax.com>,
> Paolo Amoroso <·······@mclink.it> wrote:
> >Major crash and burn of all Lisp-based shops? If by "shops" Andrei refers
> >to Lisp tools vendors, some major early Lisp shops are actually still in
> >business:
>
> And many dot-coms are still in business, too. But that doesn't contradict
> the fact that the industry crashed a year ago. If an industry is decimated
> (the term comes from a root meaning "one-tenth", i.e. reduced to 1/10th its
> size), that's a major crash and burn.
As Tim has pointed out it actually originally meant to reduce by 1/10th. (I
think 1 in 10 soldiers were executed in a mutinous Roman legion or
something like that.)
> And the companies that weren't totally wiped out were hurt badly. They're
> not as big, or they had to branch out into other businesses.
>
> In the case of Harlequin (now Xanalys), Lisp wasn't even their main product
> to begin with.
Actually Lisp was the main product to begin with. Postscript came later.
__Jason
Jason Trenouth <·····@harlequin.com> writes:
>
> Actually Lisp was the main product to begin with. Postscript came later.
>
However I get the strong impression that at the time things went wrong
the lisp part of the company was a relatively small (and I suspect
still profitable) component of the whole. This may be wrong though...
--tim
On 05 Jun 2001 22:56:06 +0100, Tim Bradshaw <···@tfeb.org> wrote:
> Jason Trenouth <·····@harlequin.com> writes:
>
> >
> > Actually Lisp was the main product to begin with. Postscript came later.
> >
>
> However I get the strong impression that at the time things went wrong
> the lisp part of the company was a relatively small (and I suspect
> still profitable) component of the whole. This may be wrong though...
If by "went wrong" you mean the "AI Winter" of the late 80s and early 90s
then the Harlequin Lisp group actually grew substantially during this time
reaching nearly 20 developers by the mid 90s. (And not including people
using Lisp to implement applications like Watson.)
However, if by "went wrong" you mean Harlequin's bankruptcy of a couple of
years ago, then yes the Lisp group had shrunk substantially by then.
Revenue-wise Lisp always brought in an order of magnitude less money than
Postscript. The printing money was invested in other printing and languages
R&D projects (and offices) that didn't bear fruit for one reason or
another.
__Jason