As was brought up recently in the in Arc thread, Common Lisp was
developed as a successor to MACLisp. However, I seem to remember
reading that there were some (perhaps half-hearted) attempts to
include InterLisp people in the standardization process. Did the
designers of Common Lisp know that if the language was successful, it
would kill off InterLisp, or was that an (un)happy accident?
Matt
In article
<····································@s19g2000prg.googlegroups.com>,
akopa <··················@gmail.com> wrote:
> As was brought up recently in the in Arc thread, Common Lisp was
> developed as a successor to MACLisp. However, I seem to remember
> reading that there were some (perhaps half-hearted) attempts to
> include InterLisp people in the standardization process. Did the
> designers of Common Lisp know that if the language was successful, it
> would kill off InterLisp, or was that an (un)happy accident?
>
> Matt
I wasn't there, so this is only from the outside:
It sounds a bit dramatic. Did the designers of the car know
that they would kill the carriage? Did the designers of
the house know that they would kill the tent and the cave?
Common Lisp was a huge investment in a successor to MacLisp.
Interlisp did not get such a large investment. There was certainly
some competition. One can see that Xerox was the most
active Interlisp vendor (AFAIK) and
sometimes Xerox was not really successful bringing their
technology (here the InterLisp-D system) to the market.
Interlisp was also running on other machines and it was
a huge language with lots of features.
I have this book: Interlisp, the language and its
usage. It is 1180 pages (by Stephen H. Kaisler).
I didn't really see Interlisp being used much here in
Europe. It was mostly tied to Xerox' InterLisp-D systems,
which died immediately when more powerful (-> more powerful
hardware) Lisp implementations came available (and no more
powerful Interlisp implementations). There were other Interlisp
implementations - but I never have seen them used here.
The Xerox Interlisp-D then also included an
implementation of Common Lisp: Xerox Common Lisp.
One effect was, that Common Lisp as a language was
not developed with an Interlisp-D like environment
in mind. It might have even prevented such an
environment to be developed - in some way. At least
it was 'culturally' incompatible with an Interlisp-D
environment. It is kind of ironic that the 'MIT people'
later got the task to develop a kind of Interlisp-D
successor (Apple Dylan). Interlisp-D as an environment
was a bit different from most Lisp systems that are
used today - it can be best compared with the Smalltalk
environments. It was working more off the list structure
(by using structure editors) and providing integrated source
control.
There was competition between the 'Maclisp' community and
the Interlisp community (which I think was smaller,
less diverse and had less financial backing).
This was not only about the base Lisp, but about
machines, development environment, expert systems -
even editors. Emacs killed S-Edit - in a way.
See also here the 'Report on Common Lisp to the Interlisp Community'.
http://www.softwarepreservation.org/projects/LISP/commonlisp/Masinter-vanMelle-Report.pdf
Rainer Joswig <······@lisp.de> writes:
> It is kind of ironic that the 'MIT people' later got the task to
> develop a kind of Interlisp-D successor (Apple Dylan).
I what way was Dylan a kind of Interlisp-D successor?
On Feb 14, 8:57 am, Lars Brinkhoff <·········@nocrew.org> wrote:
> Rainer Joswig <······@lisp.de> writes:
> > It is kind of ironic that the 'MIT people' later got the task to
> > develop a kind of Interlisp-D successor (Apple Dylan).
>
> I what way was Dylan a kind of Interlisp-D successor?
I mentioned 'Apple Dylan', because that was not 'just' the language,
but the environment from Apple. InterLisp-D is also an whole
environment and not 'just' a language. 'kind of', because both
explored how to get away from working with text files as (one of) the
primary operation for the programmer.
Apple Dylan was an environment that worked on top of a source code
database. The programmer worked with browsers and edited using those
browsers.
Say, you want to edit a function FOO. On a MIT Lisp Machine you say
(ed 'foo), the environment consults the data stored in the running
image, locates the file, opens the file in an editor and tries to find
the function in the file, the developer makes the change and saves the
file. That's the basic idea - as always, in reality it got a bit more
complicated than that.
In Apple Dylan you would browse the library or query the library to
find FOO. The environment then display then just this method for you
to edit in an in-place editor.
'Similar' to what Xerox' Smalltalk and Interlisp-D environments did.
InterLisp-D had a lot of functionality that treated the source like a
database. Then for editing, the source was treated as data and one
edited that data - not text. It tracked all changes, etc. Most people
may have read about 'DWIM' - but don't overlook the many other
interesting facilities of InterLisp-D. A good overview is here:
http://larry.masinter.net/interlisp-ieee.pdf
akopa <··················@gmail.com> writes:
> As was brought up recently in the in Arc thread, Common Lisp was
> developed as a successor to MACLisp. However, I seem to remember
> reading that there were some (perhaps half-hearted) attempts to
> include InterLisp people in the standardization process. Did the
> designers of Common Lisp know that if the language was successful, it
> would kill off InterLisp, or was that an (un)happy accident?
Mostly I think there was a clear sense that if Common Lisp was not
created, it might spell the end of MACLISP. We [the MACLISP
community] were dependent on the funding that ARPA, and ARPA was tired
of funding work where the results from one university's research
couldn't be snapped together with the results from another's and be
expected to "just work". They had proposed that a single language be
used, and my recollection is that the proposal was INTERLISP. So my
sense is that we didn't have a deathwish for INTERLISP, just a desire
to survive an apparent fight to the death destined to have only one
winner. I'll leave it to you to figure whether that's different. :)
In the community I was in, at MIT, INTERLISP was respected for some of
its accomplishments, but rarely, if ever, for its architecture, which
it seemed to me was most typically characterized as little more than a
"tower of cards". It seemed (through second-hand explanations) a
language that was about applications, not about architecture.. that
is, its internals were often sacrificed for the sake of some end goal.
Which meant it had a very sophisticated UI and some very powerful AI
tools that they liked to brag about, but nothing shiny underlying
that. I recall discussing the distinction at one point in these
terms: it's easier to build a cool UI around a solid core than a solid
core around a cool UI. DWIM was one of the big things, though there
were others.
Btw, I never personally used INTERLISP, so my knowledge of it comes
from having once read its (rather large) manual, and from having
talked to various (probably biased) refugees of INTERLISP--often about
DWIM and its various associated "horror stories" (tales of good
intentions gone awry)... Oh, and, I did once also work, in about 1978
(my first telecommuting job), with one person (Dave Barstow) who was
forced to use MACLISP and seemed to hate it and immediately set to
terraforming it to make it seem like it was INTERLISP. Through the
process of watching this, I learned that some people genuinely adored
INTERLISP and that the "INTERLISP Way" was VERY different than the
"MACLISP Way".
(Disclaimer: As an undergrad at MIT, I worked on MACSYMA at the Lab
for Computer Science at MIT, and, as fate would have it, was assigned
a desk in the relatively crowded little office where Guy Steele and
JonL White also sat. I was fully integrated into the technical side
of things, but only had second-hand knowledge of the whole big funding
issue that drove this whole thing. I wasn't involved in the founding
of the committee that made CLTL--I joined that project somewhat by
accident about a year after it started, so I didn't sit in any
meetings about "purpose". My knowledge came mostly from occasional
conversations I happened to stumble into with Steele and others around
"the Lab".)
But I double-checked what I know against the "uncut" version of "The
Evolution of Lisp" at http://www.dreamsongs.com/Files/HOPL2-Uncut.pdf
and it seems to confirm some of what I had to say above when on
page 36, it says:
| 2.10 Early Common Lisp
|
| If there were no consolidation in the Lisp community at
| this point, Lisp might have died. ARPA was not interested
| in funding a variety of needlessly competing and gratuitously
| different Lisp projects. And there was no commercial arena--yet.
I suppose it's a good thing there was not yet a Java. Then again,
Java might not have come to pass without Lisp, since some of its
founders came from the Lisp community. And also, the speed hit one
took in early Java would have probably killed it early on, just as
performance was a constant barrier for Lisp acceptance at that time.
Lisp's having persisted and worked on how to overcome some of those
barriers probably, again, contributed to Java's later successes.
Timing is everything.
A footnote about the "killed" Interlisp community: They rejoined the
fold, so to speak, during the design of ANSI Common Lisp. I think
that was sort of the tacit acknowledgment that INTERLISP was not the
way forward. It's a shame. I don't think any of us wanted INTERLISP
to go away. We just couldn't afford to have our community destroyed
in an attempt to save them.
Kent M Pitman wrote:
> akopa <··················@gmail.com> writes:
>
>> As was brought up recently in the in Arc thread, Common Lisp was
>> developed as a successor to MACLisp. However, I seem to remember
>> reading that there were some (perhaps half-hearted) attempts to
>> include InterLisp people in the standardization process. Did the
>> designers of Common Lisp know that if the language was successful, it
>> would kill off InterLisp, or was that an (un)happy accident?
>
> Mostly I think there was a clear sense that if Common Lisp was not
> created, it might spell the end of MACLISP. We [the MACLISP
> community] were dependent on the funding that ARPA, and ARPA was tired
> of funding work where the results from one university's research
> couldn't be snapped together with the results from another's and be
> expected to "just work". They had proposed that a single language be
> used, and my recollection is that the proposal was INTERLISP. So my
> sense is that we didn't have a deathwish for INTERLISP, just a desire
> to survive an apparent fight to the death destined to have only one
> winner. I'll leave it to you to figure whether that's different. :)
[snip!]
> A footnote about the "killed" Interlisp community: They rejoined the
> fold, so to speak, during the design of ANSI Common Lisp. I think
> that was sort of the tacit acknowledgment that INTERLISP was not the
> way forward. It's a shame. I don't think any of us wanted INTERLISP
> to go away. We just couldn't afford to have our community destroyed
> in an attempt to save them.
In addition to concerns that projects couldn't be hooked together
readily when built on different Lisps (and it wasn't just Maclisp and
Interlisp - there were at least UCI Lisp and Portable Standard Lisp as
lesser players), ARPA in particular and DoD generally (I worked for the
Navy at the time) had concerns about hardware portability. The DEC
Jupiter project (successor to the KL-10) melted down, making a migration
to a new architecture a near-term requirement. ARPA funded an effort to
port Interlisp to the VAX, which took a long time and many dollars, and
produced a fairly marginal initial product. It was clear that ports onto
a wide variety of commercial architectures would be a losing battle - at
just about the time that rapid evolution in the microprocessor world was
generating a host of alternatives and the Unix/C workstation was being
birthed. (And those workstations in turn killed off the custom hardware
approach from Symbolics, LMI, TI, and Xerox, which ARPA had also
invested in.)
akopa wrote:
> As was brought up recently in the in Arc thread, Common Lisp was
> developed as a successor to MACLisp. However, I seem to remember
> reading that there were some (perhaps half-hearted) attempts to
> include InterLisp people in the standardization process. Did the
> designers of Common Lisp know that if the language was successful, it
> would kill off InterLisp, or was that an (un)happy accident?
Richard Gabriel and Guy Steele have written a pretty thorough account of
the history of Lisp, including Common Lisp, and I think they cover that
part of the history quite well. See "The Evolution of Lisp" at
http://www.dreamsongs.com/Essays.html
Pascal
--
1st European Lisp Symposium (ELS'08)
http://prog.vub.ac.be/~pcostanza/els08/
My website: http://p-cos.net
Common Lisp Document Repository: http://cdr.eurolisp.org
Closer to MOP & ContextL: http://common-lisp.net/project/closer/
akopa wrote:
> As was brought up recently in the in Arc thread, Common Lisp was
> developed as a successor to MACLisp. However, I seem to remember
> reading that there were some (perhaps half-hearted) attempts to
> include InterLisp people in the standardization process. Did the
> designers of Common Lisp know that if the language was successful, it
> would kill off InterLisp, or was that an (un)happy accident?
>
> Matt
Adding to what has already been said: yes, we definitely
invited InterLisp people to participate, and Larry
Mastinter did participate. However, I think it was quite
clear to everyone that DARPA wanted to standardize on some
Lisp dialect, and either it was going to be Interlisp,
or all the post-MacLisp guys were going to get together
and make one "common" dialect that would be an alternative
to InterLisp.
So we didn't particularly try to "kill off" InterLisp, but
DARPA had set up what amounted to a competition, and if we
lost, we would have to use InterLisp ourselves, a prospect
that we did not find appetizing. We liked what we had and
wanted to keep using it. And we all knew that it would be
good for all of us to get together and standardize, but
DARPA's push for standardization is what it took to get
everybody to be willing to compromise their ideals and
favorite language notions in order to truly achieve a
synthesis.
-- Dan