Wikipedia says:
"In 1987, Interlisp was ported to the Sun Microsystems SPARC 4
architecture by a team at Xerox AI Systems (XAIS) in Sunnyvale"
Anyone know what happened?
TIA
--
Program A uses CLOS, Program B is implemented with structs, leading
to a fourfold increase in execution speed. --J. B. Heimatseiten
"GP lisper" <········@CloudDancer.com> wrote in message
················································@teranews...
> Wikipedia says:
>
> "In 1987, Interlisp was ported to the Sun Microsystems SPARC 4
> architecture by a team at Xerox AI Systems (XAIS) in Sunnyvale"
>
> Anyone know what happened?
search - http://groups-beta.google.com/group/comp.lang.lisp
for - lfg sparc
LFG is some grammer / parser demo, you'd think I'd pay more attention,
having INFORM related projects recently...
in Medley which still has a web page - http://top2bottom.net/venue.html, a
big display helps
Hmm, since the death of HyperCard, wonder how much interest in a CL
NoteCards could there be?
On Tue, 23 Aug 2005 02:55:09 GMT, <·····@yahoo.com> wrote:
>
>
>
> "GP lisper" <········@CloudDancer.com> wrote in message
> ················································@teranews...
>> Wikipedia says:
>>
>> "In 1987, Interlisp was ported to the Sun Microsystems SPARC 4
>> architecture by a team at Xerox AI Systems (XAIS) in Sunnyvale"
>>
>> Anyone know what happened?
>
> search - http://groups-beta.google.com/group/comp.lang.lisp
> for - lfg sparc
>
> LFG is some grammer / parser demo
Yes, I have seen that. It's an application, I'm interested in the
Interlisp port to Solaris (??).
--
Program A uses CLOS, Program B is implemented with structs, leading
to a fourfold increase in execution speed. --J. B. Heimatseiten
GP lisper wrote:
>
> Yes, I have seen that. It's an application, I'm interested in the
> Interlisp port to Solaris (??).
>
The application *was* the port. What they did was write some support
code which provided whatever was needed to run the InterLisp
microcode, and then you just ran your normal sysout with its output
appearing in an X window (I think - it just might have used the native
screen directly but I don't think so). It gave you an emulated D
machine running on a Sun (at about D machine speeds) - you could quite
literally run the sysout from a D machine on it. It ran on SunOS 4
(before it was retrospectively renamed Solaris 1), and I guess later
systems as well.
--tim
On 23 Aug 2005 02:30:06 -0700, <··········@tfeb.org> wrote:
>
>
> GP lisper wrote:
>
>>
>> Yes, I have seen that. It's an application, I'm interested in the
>> Interlisp port to Solaris (??).
>>
>
> The application *was* the port. What they did was write some support
> code which provided whatever was needed to run the InterLisp
> microcode, and then you just ran your normal sysout with its output
> appearing in an X window (I think - it just might have used the native
> screen directly but I don't think so). It gave you an emulated D
> machine running on a Sun (at about D machine speeds) - you could quite
> literally run the sysout from a D machine on it. It ran on SunOS 4
> (before it was retrospectively renamed Solaris 1), and I guess later
> systems as well.
The README at ftp://ftp.parc.xerox.com/pub/lfg doesn't sound like
that, it has always sounded like an application. So the sysout is a
full Interlisp image? Not a partial image (better than nothing I
suppose)?
That README indicates that the code can run in linux. If it was an
Interlisp, I would expect to hear more about it. At ILC, there was a
lot of talk about the differences in program storage between Interlisp
and current methods. If an Interlisp image was easily available, I
would have expected someone to point that out.
Looking further into LFG in any case....
--
Program A uses CLOS, Program B is implemented with structs, leading
to a fourfold increase in execution speed. --J. B. Heimatseiten
GP lisper wrote:
>
> The README at ftp://ftp.parc.xerox.com/pub/lfg doesn't sound like
> that, it has always sounded like an application. So the sysout is a
> full Interlisp image? Not a partial image (better than nothing I
> suppose)?
I don't remember if you could do partial sysouts (images) in
InterLisp-D. I think you maybe could, but they were things which
needed to load on top of an existing sysout (I seem to remember that
you could keep the plain sysout on a fileserver, then the incremental
ones would be on the machine's local disk). Anyway, I'm fairly sure
it's a complete sysout.
>
> That README indicates that the code can run in linux. If it was an
> Interlisp, I would expect to hear more about it. At ILC, there was a
> lot of talk about the differences in program storage between Interlisp
> and current methods. If an Interlisp image was easily available, I
> would have expected someone to point that out.
The later InterLisp-D versions (medley for sure, and I think lyric and
may be earlier) had Common Lisp and InterLisp in the same sysout (the
Common Lisp didn't have CLOS though - you could get PCL to run but it
was pretty slow).
I have no idea why anyone would care about running InterLisp any more.
On 24 Aug 2005 04:52:01 -0700, <··········@tfeb.org> wrote:
>
> I have no idea why anyone would care about running InterLisp any more.
I didn't follow the discussion completely, something to do with the
way that the editor stored code. It was not as source code, but some
special format that had some special advantage, all opaque to me alas.
--
Program A uses CLOS, Program B is implemented with structs, leading
to a fourfold increase in execution speed. --J. B. Heimatseiten
GP lisper wrote:
> I didn't follow the discussion completely, something to do with the
> way that the editor stored code. It was not as source code, but some
> special format that had some special advantage, all opaque to me alas.
Yeah, files[*] weren't what you might think they were in Interlisp:
they were really big long lists with the source of your program
represented as s-expressions with some special deviousness for comments
and so forth (I think that comments were (* "blah" ...), where * was
the Interlisp symbol *, IL:* not CL:* in CL terms. Really old comments
might have not even been strings but something like (COMMENT THIS
FUNCTION DOES NOT WORK). Everything was then edited using a structure
editor - DEdit was the old horrible one and SEdit was the nice one,
which was almost usable (in the sense that it was immediately apparent
that it was the coolest thing in the world, and only later became
apparent that you'd now wasted 9 billion hours on it and actually vi
would just have been better, or in fact ed would have been better). It
was all a nice idea but full of things that didn't work right, and not
CL friendly (how do you do #+ and #- for instance: answer, well, you
don't).
Like the rest of the system it was basically a huge bug with some
really cool features. The danger was that if you survived the first
few days (if anyone unused to them tried to use them they would
*immediately* crash in the most horrible way (if you were *lucky*
they'd just lock up, with a little error code on the front panel which
you could translate but generally turned out to mean `I'm fucked', if
you weren't you'd enter some reference-counted hell which would destroy
the machine over a period of hours), you developed this huge blind spot
about how awful it all was and the features started to look
interesting. Then you were basically doomed: various institutions are
still full of the twitching and dribbling shells of Interlisp users.
There used to be hope that they could be released into the community,
but I think that's gone now, after what happened. Death will be a
mercy to them.
--tim
[* There was, or could be, a filesystem in the more conventional sense
as well, with truly astonishing performance. I remember waiting to
copy some big (hah, big, right) file and thinking that it was really
getting on now because it was sustaining what I now remember as 30k/sec
but might actually have been 3k/sec, and realising that I could do way
better than this to a floppy on a proper machine.]
On 24 Aug 2005 13:23:07 -0700, <··········@tfeb.org> wrote:
>
> Like the rest of the system it was basically a huge bug with some
> really cool features.
So is W32, but that never stopped it much. Thanks for the info.
--
Program A uses CLOS, Program B is implemented with structs, leading
to a fourfold increase in execution speed. --J. B. Heimatseiten
GP lisper wrote:
> On 24 Aug 2005 13:23:07 -0700, <··········@tfeb.org> wrote:
> >
> > Like the rest of the system it was basically a huge bug with some
> > really cool features.
>
> So is W32, but that never stopped it much. Thanks for the info.
Windows is just enormously more stable & reliable than the dmachines
were. Of course, when they were current everything was less reliable
than it is now, but even by the standards of the time they were flaky
and weird.
--tim
"Tim Bradshaw" <··········@tfeb.org> writes:
> Like the rest of the system it was basically a huge bug with some
> really cool features.
Did you use these machines stand-alone? We had a few stand-alone ones,
and I got them to work reasonably stable (well, tcp/ip only worked for
one day, so I never got to use my screenshot-to-postscript interlisp
program the way I hoped for :(, but at least PUP worked ;)), but when
I worked as a summer intern at Xerox PARC in 1988, I discovered that
they were much, much more usable in connection with file servers. I
think the main problem with running them stand-alone was that
single-machine setups simply weren't properly supported by Xerox.
--
(espen)
Espen Vestre wrote:
> Did you use these machines stand-alone? We had a few stand-alone ones,
> and I got them to work reasonably stable (well, tcp/ip only worked for
> one day, so I never got to use my screenshot-to-postscript interlisp
> program the way I hoped for :(, but at least PUP worked ;)), but when
> I worked as a summer intern at Xerox PARC in 1988, I discovered that
> they were much, much more usable in connection with file servers.
Both. We had a fairly big setup with a couple of fileservers and a
print server and 20-30 dmachines. (1109s (dandytigers) and 1186s
(doves, I think) - the 1109s were splendidly antique but basically
great, the 1186s were crap - horrible unreliable screen and the rest of
the HW wasn't much better. And of course the 1109s, being older, were
faster than the 1186s in the normal Xerox way.) After all that died I
kept a single 1186 (sweet, it was called, it was mysteriously about 10%
faster than all the others) alive for a year or so, standalone.
You're right, they were better in their network (especially the 1109s
which I don't think had big enough disks to run standalone really), but
they were still basically awful. Of course there were things done at
PARC which never made it outside, including much faster hardware, and
also people who really understood them and could make them do wonderful
things.
--tim
Tim Bradshaw wrote:
> You're right, they were better in their network (especially the 1109s
> which I don't think had big enough disks to run standalone really), but
> they were still basically awful. Of course there were things done at
> PARC which never made it outside, including much faster hardware, and
> also people who really understood them and could make them do wonderful
> things.
Do you think there are ideas in Interlisp that would be worthwhile to
rediscover/reevaluate/whatever, or do you think that it's not worth it
at all?
Pascal
--
OOPSLA'05 tutorial on generic functions & the CLOS Metaobject Protocol
++++ see http://p-cos.net/oopsla05-tutorial.html for more details ++++
Pascal Costanza wrote:
> Do you think there are ideas in Interlisp that would be worthwhile to
> rediscover/reevaluate/whatever, or do you think that it's not worth it
> at all?
SEDIT was very nice.
Paul
Paul Dietz wrote:
> Pascal Costanza wrote:
>
> > Do you think there are ideas in Interlisp that would be worthwhile to
> > rediscover/reevaluate/whatever, or do you think that it's not worth it
> > at all?
>
> SEDIT was very nice.
>
SEDIT and its successor DEdit (which knew about Common Lisp
conventions) were OK for editing code, but where they really shone was
in the inspector and debugger. You could select a variable in a stack
frame, pop up a DEdit window containing its value, and mess with it any
way you wanted to, knowing you were modifying the by-ghod exact
structure your program was going to see when you continued execution.
Something like DEdit ought to be possible in CLIM - I have vague
memories of DEdit's internals talking about presentations.
Pascal Costanza wrote:
>
> Do you think there are ideas in Interlisp that would be worthwhile to
> rediscover/reevaluate/whatever, or do you think that it's not worth it
> at all?
I don't know: there might be. I do know that *I* don't want to spend
time doing so (but oh, I just have done, damn). However that's a bit
like someone who (say) fought in the great war not really wanting to
think about it too much - it doesn't mean it is not historically
interesting or that there are not lessons to be learnt from it. On the
other hand I think that the Lisp community is (or was) inclined to
romanticize certain now-dead environments far beyond anything they
justify: in my opinion that effort would better be spent looking
forwards rather than backwards.
--tim
From: Michael J. Ferrador
Subject: Re: LFG Medley Re: SPARC Interlisp?
Date:
Message-ID: <v7FPe.14$cY.10@trndny06>
"Tim Bradshaw" <··········@tfeb.org> wrote in message
·····························@g43g2000cwa.googlegroups.com...
> GP lisper wrote:
>
> > I didn't follow the discussion completely, something to do with the
> > way that the editor stored code. It was not as source code, but some
> > special format that had some special advantage, all opaque to me alas.
>
> Yeah, files[*] weren't what you might think they were in Interlisp:
> -- blah, blah -- special deviousness for comments --
> -- coolest thing in the world, and only later became apparent that
> you'd now wasted 9 billion hours on it and actually vi would be better,
> or in fact ed would have been better). -- It was all a nice idea but full
> of things that didn't work right, and not CL friendly (how do you do
#+ and #- for instance: answer, well, you don't).
> --
> Like the rest of the system it was basically a huge bug with some
> really cool features. The danger was that if you survived the first
> few days (if anyone unused to them tried to use them they would
> *immediately* crash in the most horrible way (if you were *lucky*
> they'd just lock up, with a little error code on the front panel which
> you could translate but generally turned out to mean `I'm fucked', if
> you weren't you'd enter some reference-counted hell which would destroy
> the machine over a period of hours), you developed this huge blind spot
> about how awful it all was and the features started to look
> interesting. Then you were basically doomed: various institutions are
> still full of the twitching and dribbling shells of Interlisp users.
> There used to be hope that they could be released into the community,
> but I think that's gone now, after what happened. Death will be a
> mercy to them.
So was InterLisp buggy, the Hardware/Microcode, -or- Both ?
(pulls open his Smalltalk-80 Green book, Dorado chapter)
Darn, I'm guessing the Medley emulator is "running" InterLisp-D microcode
So no smalltalk sysouts I'm guessing, (are any archived ?) well there's
always squeak.
Well of course we would not be testing actual hardware, but the stability
of a simulated Microcode and Environment.
"Tim Bradshaw" <··········@tfeb.org> writes:
> The later InterLisp-D versions (medley for sure, and I think lyric and
> may be earlier) had Common Lisp and InterLisp in the same sysout (the
> Common Lisp didn't have CLOS though - you could get PCL to run but it
> was pretty slow).
I'm pretty sure that lyric was the first "XCL" version - and it was
still pretty buggy.
> I have no idea why anyone would care about running InterLisp any more.
You probably don't understand archeologists either ;)
--
(espen)
Espen Vestre wrote:
>
> You probably don't understand archeologists either ;)
Archeological interest is a good thing, but Lisp people spend *way* too
much time obsessing about the past already.
--tim
GP lisper <········@CloudDancer.com> writes:
> The README at ftp://ftp.parc.xerox.com/pub/lfg doesn't sound like
> that, it has always sounded like an application. So the sysout is a
> full Interlisp image? Not a partial image (better than nothing I
> suppose)?
AFAIR, it's not only an Interlisp image but an Xerox Common
Lisp, i.e. it includes both Interlisp and CL.
--
(espen)
GP lisper <········@CloudDancer.com> writes:
> Wikipedia says:
>
> "In 1987, Interlisp was ported to the Sun Microsystems SPARC 4
> architecture by a team at Xerox AI Systems (XAIS) in Sunnyvale"
>
> Anyone know what happened?
This may be the product you are interested in:
Medley
http://top2bottom.net/venue.html
According to the web site, it is available for SPARC under Solaris 5
and 6 (may also work under 7, 8, or 9).
Paolo
--
Why Lisp? http://lisp.tech.coop/RtL%20Highlight%20Film
Recommended Common Lisp libraries/tools:
- ASDF/ASDF-INSTALL: system building/installation
- CL-PPCRE: regular expressions
- UFFI: Foreign Function Interface