From: Peter Seibel
Subject: Lisp and the early history of the Internet?
Date: 
Message-ID: <m365pcdbl5.fsf@javamonkey.com>
Last night I read the Steele/Gabriel essay, The Evolution of Lisp, in
HOPL2. (Also available at the URL Erann Gat posted a few days ago[1].)
Fascinating stuff.

While mulling that over I realized that there's one area of Lisp
history that I haven't seen discussed here or, really, anywhere: the
relation of the Lisp to the early development of the internet.

Given the ARPA and BBN involvement in the ARPANET *and* Lisp I'd
expect there'd be some overlap. And the Lisp boomtimes would, if I'm
not screwing up the timelines, overlap quite a bit with the
development of the internet.

Yet somehow, Unix ended up the defacto OS of the net--everybody uses
"BSD-style" sockets and folks credit the availability of the BSD
TCP/IP stack for the spread of TCP/IP networking.

I'm sure some of the folks on this list have some knowledege about
this history. Were the Lisper's too busy doing Lisp things to be
involved in the design of the internet? Or if they were involved, why
is the net so Unix-centric today?[2]

-Peter

[1] <http://www.dreamsongs.com/NewFiles/Hopl2.pdf>

[2] For example, none of the canonical internet services, sendmail,
bind, telnet, etc. are written in Lisp or (as far as I know) ever
were.


-- 
Peter Seibel                                      ·····@javamonkey.com

  The intellectual level needed   for  system design is  in  general
  grossly  underestimated. I am  convinced  more than ever that this
  type of work is very difficult and that every effort to do it with
  other than the best people is doomed to either failure or moderate
  success at enormous expense. --Edsger Dijkstra

From: Gabe Garza
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <87lly8eif4.fsf@ix.netcom.com>
Peter Seibel <·····@javamonkey.com> writes:
> [2] For example, none of the canonical internet services, sendmail,
> bind, telnet, etc. are written in Lisp or (as far as I know) ever
> were.

I'm really hoping the Wise Elders will chime in and give lots of
insightful anecdotes, but there certainly were Lisp implementations of
"canonical" IP services.  For example, all Symbolics computers have
Lisp IP daemons:

······@bob:~$ telnet ivory-tower 25
Trying 192.168.1.4...
Connected to 192.168.1.4.
Escape character is '^]'.
220  kynopolis.org SMTP service ready.
HELP
214- Symbolics Lisp Machine SMTP server; Zmail 437.0.
214- Protocol is RFC 821.
214  Commands supported are HELO, MAIL, RCPT, DATA, RSET, SEND, SOML, 
     VRFY, EXPN, HELP, NOOP, QUIT, and XCTM.

······@bob:~$ ftp ivory-tower
Connected to ivory-tower.
220  FTP service ready.
Name (ivory-tower:ggarza): GNG
230  USER accepted.
Remote system type is .
ftp> ls *.lisp
200  PORT 192,168,1,1,138,186 (BOB port 35514) accepted.
150  Here comes a listing of >GNG>*.lisp.newest.
>GNG>gabriel-benchmarks.lisp.1  16  77333  12/18/02 22:39:53  Lisp-Machine
>GNG>lispm-init.lisp.11  1  238  1/04/03 14:55:03  GNG
226  File transfer completed.
ftp> 

There's also telnet, finger, and more...

C is the language of Unix, so the IP daemons on Unix were written in
C; Lisp is the language of Lisp Machines, so the daemons were in Lisp;
etc...

Gabe Garza
From: Scott McKay
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <H9Jna.497994$L1.143936@sccrnsc02>
"Gabe Garza" <·······@ix.netcom.com> wrote in message
···················@ix.netcom.com...
> Peter Seibel <·····@javamonkey.com> writes:
> > [2] For example, none of the canonical internet services, sendmail,
> > bind, telnet, etc. are written in Lisp or (as far as I know) ever
> > were.
>
> I'm really hoping the Wise Elders will chime in and give lots of
> insightful anecdotes, but there certainly were Lisp implementations of
> "canonical" IP services.  For example, all Symbolics computers have
> Lisp IP daemons:

I myself wrote some of this.

Most all of the "important stuff" as of, say, 1988, was implemented.
I can't remember if we did NFS, though.  I don't think that BIND
even existed at that point.  'sendmail' is, of course, just a Unix program
that implements mail protocols; the Lisp machine did this, too.

> ······@bob:~$ telnet ivory-tower 25
> Trying 192.168.1.4...
> Connected to 192.168.1.4.
> Escape character is '^]'.
> 220  kynopolis.org SMTP service ready.
> HELP
> 214- Symbolics Lisp Machine SMTP server; Zmail 437.0.
> 214- Protocol is RFC 821.
> 214  Commands supported are HELO, MAIL, RCPT, DATA, RSET, SEND, SOML,
>      VRFY, EXPN, HELP, NOOP, QUIT, and XCTM.
>
> ······@bob:~$ ftp ivory-tower
> Connected to ivory-tower.
> 220  FTP service ready.
> Name (ivory-tower:ggarza): GNG
> 230  USER accepted.
> Remote system type is .
> ftp> ls *.lisp
> 200  PORT 192,168,1,1,138,186 (BOB port 35514) accepted.
> 150  Here comes a listing of >GNG>*.lisp.newest.
> >GNG>gabriel-benchmarks.lisp.1  16  77333  12/18/02 22:39:53  Lisp-Machine
> >GNG>lispm-init.lisp.11  1  238  1/04/03 14:55:03  GNG
> 226  File transfer completed.
> ftp>
>
> There's also telnet, finger, and more...
>
> C is the language of Unix, so the IP daemons on Unix were written in
> C; Lisp is the language of Lisp Machines, so the daemons were in Lisp;
> etc...
>
> Gabe Garza
From: Christopher C. Stacy
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <ud6jkmnip.fsf@dtpq.com>
>>>>> On Fri, 18 Apr 2003 02:11:51 GMT, Scott McKay ("Scott") writes:

 Scott> "Gabe Garza" <·······@ix.netcom.com> wrote in message
 Scott> ···················@ix.netcom.com...
 >> Peter Seibel <·····@javamonkey.com> writes:
 >> > [2] For example, none of the canonical internet services, sendmail,
 >> > bind, telnet, etc. are written in Lisp or (as far as I know) ever
 >> > were.
 >> 
 >> I'm really hoping the Wise Elders will chime in and give lots of
 >> insightful anecdotes, but there certainly were Lisp implementations of
 >> "canonical" IP services.  For example, all Symbolics computers have
 >> Lisp IP daemons:

 Scott> I myself wrote some of this.

 Scott> Most all of the "important stuff" as of, say, 1988, was implemented.
 Scott> I can't remember if we did NFS, though.

Yes.
From: Will Hartung
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <gqLna.503$hu.62547218@newssvr15.news.prodigy.com>
"Peter Seibel" <·····@javamonkey.com> wrote in message
···················@javamonkey.com...
> All I'm really saying is:
>
>  a) Given the organizations that were involved in both the early
>  Internet *and* Lisp I'd expect there to be lots of overlap.
>
>  b) By the mid 90's, the internet was waxing, Lisp was waning, and
>  when folks wanted to get on the net the first thing they did was get
>  a Unix box--there were no signs of Lisp being in any way central to
>  the net nor even--as far as I can tell--any signs of Lisp in the
>  "fossil record" of the net.

I guess you can safely say that most of the overlap was simply done in the
interoperation of mahcines at the protocol level, so in many ways the
overlap is transparent.

Did Lisp machines ever really cooperate in the early UUCP network?

Once BSD started flourishing, it sort of became a defacto standard upon
which the Internet was based and if nothing else survived the era, the
socket API did. Yet, even today, we don't have a common socket API in CL.

I also am not aware of any of the main Internet services floating around in
Lisp source form. No doubt they were there (as was earlier demonstrated),
but they didn't survive off of the Lisp Machines. We have new versions, but
not legacy versions.

It seems to me that because these were more "Services" rather than
"applications", that it simply wasn't important that they do survive. Who
cares how mail is handled, as long as it's done properly?

Once off of the Lisp Machines, Lisp became second cousin and merely a guest
in the computer world rather than a home owner. So, while applications
survived, and Lisp utilities, basic services were left to the wayside as
they weren't really necessary, becoming part of the overarching OS
infrastructure.

Regards,

Will Hartung
(·····@msoft.com)
From: Scott McKay
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <XEdoa.525686$3D1.287856@sccrnsc01>
"Scott McKay" <···@attbi.com> wrote in message
···························@sccrnsc02...
>
> "Gabe Garza" <·······@ix.netcom.com> wrote in message
> ···················@ix.netcom.com...
> > Peter Seibel <·····@javamonkey.com> writes:
> > > [2] For example, none of the canonical internet services, sendmail,
> > > bind, telnet, etc. are written in Lisp or (as far as I know) ever
> > > were.
> >
> > I'm really hoping the Wise Elders will chime in and give lots of
> > insightful anecdotes, but there certainly were Lisp implementations of
> > "canonical" IP services.  For example, all Symbolics computers have
> > Lisp IP daemons:
>
> I myself wrote some of this.
>
> Most all of the "important stuff" as of, say, 1988, was implemented.
> I can't remember if we did NFS, though.  I don't think that BIND
> even existed at that point.  'sendmail' is, of course, just a Unix program
> that implements mail protocols; the Lisp machine did this, too.

Kent Pitman reminded me of something I had complete forgotten
w.r.t. NFS on the Lisp Machine.  He says,

   Geez, Scott, don't you remember the famous variable
   *check-passwords-for-show* when the appalled Lisp
   Machine designers found that the NFS password checking
   was administered CLIENT-SIDE (on some bogus assumptions
   that only unixes talk to other unixes, and that roots everywhere
   are trustworthy, and ...)

I guess this kludge was so awful, it caused me to forget about
NFS entirely.  I would still like to forget about NFS entirely,
but it still freezes everybody from doing work about once a
month at my present job, a form of lossage that has been there
since day zero when NFS is used in shared auto-mind file systems.
Grrrr...
From: Gorbag
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <BACB5BE3.3613%gorbagNOSPAM@NOSPAMmac.com>
On 4/19/03 8:09 AM, in article ·······················@sccrnsc01, "Scott
McKay" <···@attbi.com> wrote:

> 
> "Scott McKay" <···@attbi.com> wrote in message
> ···························@sccrnsc02...
>> 
>> "Gabe Garza" <·······@ix.netcom.com> wrote in message
>> ···················@ix.netcom.com...
>>> Peter Seibel <·····@javamonkey.com> writes:
>>>> [2] For example, none of the canonical internet services, sendmail,
>>>> bind, telnet, etc. are written in Lisp or (as far as I know) ever
>>>> were.
>>> 
>>> I'm really hoping the Wise Elders will chime in and give lots of
>>> insightful anecdotes, but there certainly were Lisp implementations of
>>> "canonical" IP services.  For example, all Symbolics computers have
>>> Lisp IP daemons:
>> 
>> I myself wrote some of this.
>> 
>> Most all of the "important stuff" as of, say, 1988, was implemented.
>> I can't remember if we did NFS, though.  I don't think that BIND
>> even existed at that point.  'sendmail' is, of course, just a Unix program
>> that implements mail protocols; the Lisp machine did this, too.
> 
> Kent Pitman reminded me of something I had complete forgotten
> w.r.t. NFS on the Lisp Machine.  He says,
> 
>  Geez, Scott, don't you remember the famous variable
>  *check-passwords-for-show* when the appalled Lisp
>  Machine designers found that the NFS password checking
>  was administered CLIENT-SIDE (on some bogus assumptions
>  that only unixes talk to other unixes, and that roots everywhere
>  are trustworthy, and ...)

I considered it such a laughable feature of NFS, I made it a regular part of
the LispM demo at our CS department. "See, the Symbolics machine thinks that
your lack of a password to someone else's NFS mounted filesystem is really a
bug and drops into the debugger. Just press symbol-f to continue as if you
had entered their password. The server won't mind."
From: Andrew Burton
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <20030418090755.19020.00000137@mb-m13.aol.com>
My best suggest, for whatever it's worth, is to read "The Rise of Worse is
Better"(1) and then follow it up with "In Praise of Evolvable Systems"(2).

--

1. http://www.ai.mit.edu/docs/articles/good-news/subsection3.2.1.html

2. http://www.shirky.com/writings/evolve.html

Andrew Burton -- tuglyraisin at aol dot com
Felecia Station on Harvestgain
Jipi: Don't Believe everything the Net tell you.
Chip: I agree with this. My inputs are not to be trusted.
CROW: Quick, save the Robot Ashley!
From: Peter Seibel
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <m34r4vbzk9.fsf@javamonkey.com>
···········@aol.commune (Andrew Burton) writes:

> My best suggest, for whatever it's worth, is to read "The Rise of Worse is
> Better"(1) and then follow it up with "In Praise of Evolvable Systems"(2).

Thanks. I've read [1] before but hadn't seen [2]. I'm still hoping
someone who was around BBN at the right time can fill us in on what
the relation between the guys doing Lisp stuff and the guys doing
Internet stuff was.

-Peter

> 1. http://www.ai.mit.edu/docs/articles/good-news/subsection3.2.1.html
> 
> 2. http://www.shirky.com/writings/evolve.html

-- 
Peter Seibel                                      ·····@javamonkey.com

  The intellectual level needed   for  system design is  in  general
  grossly  underestimated. I am  convinced  more than ever that this
  type of work is very difficult and that every effort to do it with
  other than the best people is doomed to either failure or moderate
  success at enormous expense. --Edsger Dijkstra
From: Christopher C. Stacy
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <ubrz2caei.fsf@dtpq.com>
>>>>> On Thu, 17 Apr 2003 22:08:01 GMT, Peter Seibel ("Peter") writes:
 Peter> I'm sure some of the folks on this list have some knowledege
 Peter> about this history. Were the Lisper's too busy doing Lisp
 Peter> things to be involved in the design of the internet? Or if
 Peter> they were involved, why is the net so Unix-centric today?[2]

First of all, I would not say that the Internet is "Unix-centric".
But I think the answer to your question is that the Lisp hackers
put all their eggs in the Lisp Machine basket.
From: Christopher Browne
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <b7sumu$43cjt$1@ID-125932.news.dfncis.de>
······@dtpq.com (Christopher C. Stacy) wrote:
>>>>>> On Thu, 17 Apr 2003 22:08:01 GMT, Peter Seibel ("Peter") writes:
>  Peter> I'm sure some of the folks on this list have some knowledege
>  Peter> about this history. Were the Lisper's too busy doing Lisp
>  Peter> things to be involved in the design of the internet? Or if
>  Peter> they were involved, why is the net so Unix-centric today?[2]

> First of all, I would not say that the Internet is "Unix-centric".

Well, Tenex, Twenex, TOPS-10/TOPS-20 and such were surely heavily
involved; anyone younger than 30 is unlikely to ever have been able to
see anything in action that was influential in Internet "innovation"
aside from Unix.

Based on "modern" standards, I'm pretty exceptional in the number of
platforms I have used; my list includes GCOS, MVS and VMS, as things
that are neither flavors of Unix nor of Windows.  Those platforms
pretty much went away as things students would have exposure to by the
'90s.

There's nothing other than Unix left to resemble "Internet
innovation."

> But I think the answer to your question is that the Lisp hackers put
> all their eggs in the Lisp Machine basket.

.. Aside from Franz, of course ...

Which is rather interesting; they /are/ still around.

You're probably quite right, though.  At the time that MIT's Lisp folk
were spinning off, the common "doctrine" was that it would be
reasonable to deploy custom hardware for the systems.  It was even the
beginning assumption of how the GNU people started out; the outlook
then involved MicroVAXen and the likes, with no belief that the "toy"
computers Intel was deploying would ever amount to anything.

The number of architectures has been pretty steadily dropping since
the '80s.  And is still dropping.

-> I see little likelihood of future MIPS architectures being 
   prominent;
-> IBM has consolidated to IA-32 and PPC (the mainframes all run
   customized versions of PPC);
-> It's not totally evident how PA-RISC and Alpha will fold into
   IA-64; we'll probably only have one of the three left, in the
   end.
-> AMD x86-64 may kill something else, perhaps IA-64.

There's enough new technology for chip design that it's not that tough
to design another one, but the problem is then in getting support for
all the necessary attendant technologies (PCI, USB, AGP being obvious
ones) so that you can actually connect hardware to the CPU.

I'll bet a group of grad students in EE could do a credible "Lisp
machine" chip design as a term project, but it would then take
*enormous* work to make it usable, and as soon as ATI or nVidia got
overcome by some new graphics chipset vendor, or Adaptec bought out
the SCSI card vendor (Mylex?  BusLogic?  Symbios/NCR?)  it would
become impossible to support the hardware anymore.  

Consider: Running Plan 9 is getting troublesome because it never has
supported Adaptec SCSI, and the chip vendors they *do* support all got
eliminated by Adaptec...
-- 
(concatenate 'string "cbbrowne" ·@acm.org")
http://cbbrowne.com/info/oses.html
Everyone has a photographic memory, some don't have film.
From: Franz Kafka
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <b3b6b110.0304200933.14c2b6e5@posting.google.com>
> 
> I'll bet a group of grad students in EE could do a credible "Lisp
> machine" chip design as a term project, but it would then take
> *enormous* work to make it usable, and as soon as ATI or nVidia got
> overcome by some new graphics chipset vendor, or Adaptec bought out
> the SCSI card vendor (Mylex?  BusLogic?  Symbios/NCR?)  it would
> become impossible to support the hardware anymore.  
> 

If the Lisp Processor was setup in a machine that supported TCP/IP,
and
was designed to be a distrubted OS that was accessable from front end
software on standard computers; the machine would always work.

The Lisp Processing unit would only take care of processing Lisp
programs
efficently--maybe Prolog, APL, etc. programs too.

The front end computer would be any computer that could send and
recieve TCP/IP sockets. (In order for it to work some software such as
CL-HTTP would need to be running on the Lisp Processor.)

The Lisp Compiler, and support software could be on EPROMS, along with
the Microcode. This way if the compiler needed to be changed the
EPROMS could be
reprogrammed.

Since the Lisp Unit would only deal with running Lisp and returning
the result over the enternet link to the users computer, the host
computer could be upgraded as long as all of the protocols in use have
some standard: IEEE, ANSI, ISO.


<<User's Computer: anymachine>> <--connector--><<Lisp Processor /w
CL-HTTP>>
Disks, Graphics,Networking, etc.                 Stack Machine/
Network Stuff
From: Christopher C. Stacy
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <ud6jhfcnb.fsf@dtpq.com>
>>>>> On 20 Apr 2003 01:55:43 GMT, Christopher Browne ("Christopher") writes:
 Christopher> .. Aside from Franz, of course ...

(I meant in terms of operating system development.)
From: Mark Hurd
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <3ea1c5a2$1_1@news.iprimus.com.au>
Peter Seibel wrote:
> Were the Lisper's too busy doing Lisp things to be
> involved in the design of the internet? Or if they were involved, why
> is the net so Unix-centric today?[2]

Searching the Google 20-year archive shows a handful of hits a month from the
earliest year in the archive (1981) for "lisp", "car cdr", "cadr", etc.

I can't be sure, but it looks like it is mostly "newbies" and references to
lists of programming languages.

Someone with more knowledge of Lisp history should try Google Groups Advanced
Search:

http://groups.google.com.au/advanced_group_search?q=car+cdr&scoring=d&as_drrb=
b&as_miny=1981&as_maxy=1984


Regards,
Mark Hurd, B.Sc.(Ma.) (Hons.)
From: Paul Wallich
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <pw-9E9457.20472219042003@reader1.panix.com>
In article <············@news.iprimus.com.au>,
 "Mark Hurd" <········@ozemail.com.au> wrote:

> Peter Seibel wrote:
> > Were the Lisper's too busy doing Lisp things to be
> > involved in the design of the internet? Or if they were involved, why
> > is the net so Unix-centric today?[2]
> 
> Searching the Google 20-year archive shows a handful of hits a month from the
> earliest year in the archive (1981) for "lisp", "car cdr", "cadr", etc.
> 
> I can't be sure, but it looks like it is mostly "newbies" and references to
> lists of programming languages.
> 
> Someone with more knowledge of Lisp history should try Google Groups Advanced
> Search:
> 
> http://groups.google.com.au/advanced_group_search?q=car+cdr&scoring=d&as_drrb=
> b&as_miny=1981&as_maxy=1984

By the nature of the early usenet you may not get a whole lot of stuff 
--  one of the reasons for building usenet was to serve people who 
didn't have good access to arpanet traffic. And most of the early 
Lispers were extremely well connected. What archives are there of the 
arpa mailing list traffic, other than random packrats' scratch tapes?

paul
From: Mark Hurd
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <3ea369cd_1@news.iprimus.com.au>
Paul Wallich wrote:
> In article <············@news.iprimus.com.au>,
>  "Mark Hurd" <········@ozemail.com.au> wrote:
> >
> > Someone with more knowledge of Lisp history should try Google Groups
> > Advanced Search:
> >
> >
http://groups.google.com.au/advanced_group_search?q=car+cdr&scoring=d&as_drrb=
b&as_miny=1981&as_maxy=1984
>
> By the nature of the early usenet you may not get a whole lot of stuff
> --  one of the reasons for building usenet was to serve people who
> didn't have good access to arpanet traffic. And most of the early
> Lispers were extremely well connected. What archives are there of the
> arpa mailing list traffic, other than random packrats' scratch tapes?

FYI this post explains the sources of the Google archive:

http://groups.google.com/groups?selm=90cbefb1.0112211746.43392bc8%40posting.go
ogle.com

Regards,
Mark Hurd, B.Sc.(Ma.) (Hons.)
From: Barry Margolin
Subject: Re: Lisp and the early history of the Internet?
Date: 
Message-ID: <OIepa.9$304.498@paloalto-snr1.gtei.net>
In article <··············@javamonkey.com>,
Peter Seibel  <·····@javamonkey.com> wrote:
>Given the ARPA and BBN involvement in the ARPANET *and* Lisp I'd
>expect there'd be some overlap. And the Lisp boomtimes would, if I'm
>not screwing up the timelines, overlap quite a bit with the
>development of the internet.

While both were going on at BBN, I suspect there wasn't that much overlap
between the two R&D groups.

One place where Lisp did seem to leave a tiny imprint was in DNS
configuration files: they use ";" as the comment character (which causes
Unix sysadmins no end of trouble -- they're so used to "#" being the
comment character in all their configuration files).  But I'm not really
sure if this is a Lisp artifact at all -- I think some assemblers also used
this, and that may be where it came from (it may also be where Lisp copied
it from).

>Yet somehow, Unix ended up the defacto OS of the net--everybody uses
>"BSD-style" sockets and folks credit the availability of the BSD
>TCP/IP stack for the spread of TCP/IP networking.

I think this is because Berkeley got a contract from the DOD to embed
TCP/IP into the Unix kernel.  This was at the same time that Unix was
spreading quickly among universities.  This combination caused Unix-based
network servers to become very popular.

-- 
Barry Margolin, ··············@level3.com
Genuity Managed Services, a Level(3) Company, Woburn, 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.