Well doggone anyway... I wanted to get a Lisp up and running on my new RH 6
Linux box and both CMUCL and CLISP bomb out before I can even say hello...
After pouring through the archives on CMUCL it appears that I have to wait
for a new build (if one ever appears) because of fundamental changes to the
Libc. CLISP, OTOH bombs with a bad file descriptor. It appears that the
CMUCL crowd went on to CMU Dylan, which is also dead from the sounds of it,
and Harlequin is also disappearing... So, doesn't anyone use Lisp on Linux?
D.McClain
"David McClain" <········@azstarnet.com> writes:
> Date: Sat, 7 Aug 1999 02:17:19 -0700
>
> Well doggone anyway... I wanted to get a Lisp up and running on my new RH 6
> Linux box and both CMUCL and CLISP bomb out before I can even say hello...
> After pouring through the archives on CMUCL it appears that I have to wait
> for a new build (if one ever appears) because of fundamental changes to the
> Libc. CLISP, OTOH bombs with a bad file descriptor. It appears that the
> CMUCL crowd went on to CMU Dylan, which is also dead from the sounds of it,
> and Harlequin is also disappearing... So, doesn't anyone use Lisp on Linux?
I suspect thos systems will all be made to work eventually, but if you are
in a hurry you can try the freepoplog system recently announced,
accessible from here
ftp://ftp.cs.bham.ac.uk/pub/dist/poplog/freepoplog.html
I am assured it runs on Intel+RedHat 6.0 though I have not tried
that combination personally.
If you fetch the linux version and you are not interested in
the other languages you can remove the sub-directories
with code and documentation for Prolog and ML, and their saved
images, to save space.
Poplog includes an incremental compiler for Common Lisp, as well as
other AI languages, as explained in previous postings to comp.lang.lisp
with the word "versions" in the Subject line.
But Poplog Common Lisp is not heavily optimised, and some things are
slower than others. The compiler and garbage collector are very
fast, but if you do lots of number crunching/array crunching it
doesn't do the kind of optimising some other lisp systems do.
E.g. at present there's no way to avoid run-time typechecking on all
floating point operations. But it is still usable for develelopment,
teaching, etc.
Since the source code is now available at the freepoplog site (by
courtesy of Sussex University) I wonder whether some Lisp experts
would be willing to investigate ways of improving the Lisp?
Questions and comments about Poplog common lisp should be posted, or
cross-posted to comp.lang.pop
Aaron
===
Aaron Sloman, ( http://www.cs.bham.ac.uk/~axs/ )
School of Computer Science, The University of Birmingham, B15 2TT, UK
EMAIL ········@cs.bham.ac.uk
Phone: +44-121-414-4775 (Sec 4774) Fax: +44-121-414-4281
--
Aaron Sloman, ( http://www.cs.bham.ac.uk/~axs/ )
School of Computer Science, The University of Birmingham, B15 2TT, UK
EMAIL A.Sloman AT cs.bham.ac.uk (NB: Anti Spam address)
PAPERS: ftp://ftp.cs.bham.ac.uk/pub/groups/cog_affect/0-INDEX.html
In article <············@soapbox.cs.bham.ac.uk>, Aaron Sloman See text
for reply address <···················@cs.bham.ac.uk> wrote:
> "David McClain" <········@azstarnet.com> writes:
>
> > Date: Sat, 7 Aug 1999 02:17:19 -0700
> >
> > Well doggone anyway... I wanted to get a Lisp up and running on my new RH 6
> > Linux box and both CMUCL and CLISP bomb out before I can even say hello...
> > After pouring through the archives on CMUCL it appears that I have to wait
> > for a new build (if one ever appears) because of fundamental changes to the
> > Libc.
I have CMUCL running on an RH 6 x686 system. There is a glibc 2.1 build
around, it's one of the "experimental" builds. The running CUMCL
advertises itself as being built on 2Feb99, I *think* it's the linux
longfloat version but couldn't be sure, I was trying out three/four
build versions at the time.
It's rather a shame that there isn't a viable Linux build-from-source
route and it is a little disappointing that it takes so long for new
builds to find their way on to the ftp site.
Still, that acts in POP's favour at the moment.
Graham Higgins <···@bel-epa.delete-this.com> writes:
> I have CMUCL running on an RH 6 x686 system. There is a glibc 2.1 build
> around, it's one of the "experimental" builds. The running CUMCL
> advertises itself as being built on 2Feb99, I *think* it's the linux
> longfloat version but couldn't be sure, I was trying out three/four
> build versions at the time.
>
> It's rather a shame that there isn't a viable Linux build-from-source
> route and it is a little disappointing that it takes so long for new
> builds to find their way on to the ftp site.
Actually, using Peter Van Eynde's Debian source packages for CMUCL you
can just type "make" and rebuild from source, but you need to have a
CMUCL binary installed. If you install his Debian package it will
compile right out of the tarball, but if you are running on RH you
might have to edit the makefile to tell it where to find your existing
core and binaries. I just built it on my P90 with 64 megs of RAM
without any problems. I have not tried using his packages for
compiling on other non-linux boxes tho.
> Still, that acts in POP's favour at the moment.
Pop does indeed seem to be easily buildable out of the box, once you
find the right sequence of commands. I have some patches to get it to
build on Debian (with no -ltermcap), and to accept #! acomments so you
can use it for shell scripting. Another thing that POP has going for
it is the excellent documentation. CMUCL regretably has very little
implmentation documentation, and so it's a nightmare trying to learn
how to rebuild, modify and maintain it, or port it.
--
Craig Brozefsky <·····@red-bean.com>
Free Scheme/Lisp Software http://www.red-bean.com/~craig
I say woe unto those who are wise in their own eyes, and yet
imprudent in 'dem outside -Sizzla
>>>>> "Craig" == Craig Brozefsky <·····@red-bean.com> writes:
Craig> it is the excellent documentation. CMUCL regretably has
Craig> very little implmentation documentation, and so it's a
Craig> nightmare trying to learn how to rebuild, modify and
Craig> maintain it, or port it.
As someone who has rebuilt and modified CMUCL, I think it would still
be a nightmare even with documentation. I understand even the main
developers have a difficult time rebuilding when something major
changes. Porting would be extremely hard whether you had
documentation or not.
I think the same could be said for gcc.[1] How much documentation is
there to help you modify, maintain, or port it?
However, as Peter Van Eynde pointed out a while ago, there are lots
of things you could do to make CMUCL better that don't require
difficult compilations. Things like improving bignum arithmetic,
making the compiler smarter, etc.
Ray
Footnotes:
[1] I have not looked at gcc, so I might be wrong.
Not following up to comp.lang.pop.
Raymond Toy <···@rtp.ericsson.se> writes:
> I think the same could be said for gcc.[1] How much documentation is
> there to help you modify, maintain, or port it?
"Porting the GNU C Compiler for Dummies"
Hans-Peter Nilsson
December 19, 1998
96 pages
ftp://ftp.axis.se/pub/users/hp/pgccfd/
...........
Maybe someone should write a "LISP for Dummies". On the other hand,
LISP is already for dummies. The language's regularity (and thus,
simplicity) makes it unnecessary to waste any brainpower on cryptic
syntax or other "warts".
(list of warty languages elided)
--
Samir Barjoud
·····@mindspring.com
On Sat, 7 Aug 1999 02:17:19 -0700, David McClain <········@azstarnet.com> wrote:
>Well doggone anyway... I wanted to get a Lisp up and running on my new RH 6
>Linux box and both CMUCL and CLISP bomb out before I can even say hello...
>After pouring through the archives on CMUCL it appears that I have to wait
>for a new build (if one ever appears) because of fundamental changes to the
>Libc. CLISP, OTOH bombs with a bad file descriptor. It appears that the
>CMUCL crowd went on to CMU Dylan, which is also dead from the sounds of it,
>and Harlequin is also disappearing... So, doesn't anyone use Lisp on Linux?
You might have more luck with Allegro CL at http://www.franz.com/ The recent
release mentions RH6 in the list of distributions they have tested on. I also got
ACL working on SuSE and someone mentioned that it worked ok on Debian too.
As for RH6, I did not upgrade because of the high probability of problems
with otherwise good solutions on RH5.2 where ACL, CMUCL, CLISP and LW
works decently which lets me test code on various implementations.
I think there was a patch for LW and RH6 the other day, and I think/hope
that the publicly available Linux offerings have been patched (there
were some obnoxious problems in the first releases which they produced
patches for but I think the "free" releases lacked the ability to
load patches).
--
------------------------------------------------------------------
Stig Erik Sandoe Institute of Informatics, University of Bergen
····@ii.uib.no http://www.ii.uib.no/~stig/
"David McClain" <········@azstarnet.com> writes:
> It appears that the CMUCL crowd went on to CMU Dylan, which is also
> dead from the sounds of it,
CMUCL Dylan became Gwydion Dylan and it is far from dead. The site for
this is http://www.gwydiondylan.org.
> So, doesn't anyone use Lisp on Linux?
Try looking for BeOS Lisp environments - you Linux people are
overflowing in lisps in comparison :-).
Chris.
"David McClain" <········@azstarnet.com> writes:
In recent times the frequency of "Is XXX Dead?" threads in c.l.l have
increased to the point that theories about general world weariness at
the end of a century/millenium seem probable again. Maybe this should
be called the "Human Millenium Bug", or shorter "HUMBUG". Or maybe it
is just the solar eclipse, and there has been too much reading of
Nostradamus... ;)
> Well doggone anyway... I wanted to get a Lisp up and running on my new RH 6
> Linux box and both CMUCL and CLISP bomb out before I can even say hello...
> After pouring through the archives on CMUCL it appears that I have to wait
> for a new build (if one ever appears) because of fundamental changes to the
> Libc. CLISP, OTOH bombs with a bad file descriptor. It appears that the
> CMUCL crowd went on to CMU Dylan, which is also dead from the sounds of it,
> and Harlequin is also disappearing... So, doesn't anyone use Lisp on Linux?
A couple of facts about CMU CL (which can be easily ascertained on
http://www.cons.org/cmucl/):
- The original developers were at CMU, where they developed CMU CL
(born from Spice Lisp) until ca. 1992, on contract from the US
government (I think it was (D)ARPA, IIRC). In 1992 they went on to
do CMU Dylan (as part of their new development environment), which
they dumped a couple of years later in favor of Java.
- Both the maintenance of CMU CL and CMU Dylan have been taken up by
different groups of volunteers (which AFAIK don't overlap at all).
For CMU Dylan look at an archive of comp.lang.dylan, which will give
you pointers to the homepage of this implementation. The loose
group of maintainers of CMU CL can be seen at the CMU CL homepage
mentioned above.
- This group of volunteers has been far from inactive. The x86 ports
(first (Free)BSD, then Linux) where done by this group, as well as
Multi-Processing support for the x86 ports. Since CMU stopped work
on CMUCL (release 17f) there have been two new releases for most
platforms (18a and 18b) supported. The Linux version has been
integrated into Debian, with more frequent releases. Of all
supported platforms, it can be said that x86 on BSD and Linux is in
best shape feature-wise, with the Sun/Sparc version in very good
shape as well. There has been recent work to bring other platforms
(DEC/Alpha and I think MIPS) up to the same level as the
Sun/Sparc version. Check the CMU CL mailing list archive at the web
site above for details.
- Summary: No, CMU CL is not dead, least of all on x86 BSD or Linux.
I can't speak for CLISP, but since I've seen the main author of CLISP
active here a couple of days ago, and he made no indication at all
that CLISP is dead (with versions from 1999 available), I don't think
that CLISP is dead either.
Furthermore Harlequin has been bought from receivership, and the new
owners have as yet made no indication of wanting to gut the whole
language department, AFAIK. The open Lispworks Personal Beta
for Linux is still available on their website, and I've received a
bug-fix for problems with Redhat 6.0/Kernel 2.2.* just yesterday. So
Harlequin Lispworks on Linux seems far from dead either.
Still furthermore, Franz' Allegro Common Lisp 5.* offerings for Linux
are still available, and I've heard nothing to indicate that this is
going away, either. And I've heard that they seem to have a version
for LinuxPPC, too (though I'm not knowledgable about availability of
this).
Even furthermore still, Eclipse, the CL implementation that compiles
to and integrates very well with C is and has been available for
Linux/Intel for some time, and it doesn't seem to go away soon,
either.
On the whole, the situation for CL on Linux/Intel seems better than
ever before, with three major commercial implementations available
now, and at least two open-source implementations available and going
strong. And even non-Intel Linux plattforms seem to be getting more
implementations (I think CLISP has supported some of them for some
time) nowadays. And that isn't counting no-longer supported
implementations like GCL & Co.
After we have cleared up the status of CL on Linux, let's turn to your
specific problem:
- There have been two "major" upheavals on the Libc front on Linux in
recent times, the transition from libc5 to glibc 2.0, and the
transition from glibc 2.0 to glibc 2.1. Since I'm not informed on
which glibc version RedHat 6.0 is using, I can't tell you which
problems you have run into. But I can assure you that CMU CL
2.4.[8-9] runs perfectly on glibc 2.0 based systems (such as Debian
2.1, of which that particular version is part). Furthermore I think
that problems with glibc 2.1 have been fixed some time ago, since
the upcoming Debian 2.2 release is based on glibc 2.1, and this will
contain CMU CL 2.4.13 or 2.4.14, I think.
- So before anything else, I'd suggest getting the right Debian
packages for your glibc (either the ones from stable => Debian 2.1
for glibc 2.0, or from unstable => upcoming Debian 2.2 for glibc
2.1) from your nearest Debian mirror (see http://www.debian.org/)
and converting them to RPM packages via alien. You can also unpack
Debian packages manually (they are ar archives, with two tarballs
inside, one with the files, the other can be ignored). The packages
are in the devel sections of each distribution.
- If this fails, mail me...
Good luck, have fun, and cheer up... ;)
Regs, Pierre.
--
Pierre Mai <····@acm.org> PGP and GPG keys at your nearest Keyserver
"One smaller motivation which, in part, stems from altruism is Microsoft-
bashing." [Microsoft memo, see http://www.opensource.org/halloween1.html]
From: ········@cc.hut.fi
Subject: Re: Is the King Dead? (was Re: Is Lisp Dead on Linux? (Far from it!))
Date:
Message-ID: <m3zp03r63i.fsf@mu.tky.hut.fi>
····@acm.org (Pierre R. Mai) writes:
> - So before anything else, I'd suggest getting the right Debian
> packages for your glibc (either the ones from stable => Debian 2.1
> for glibc 2.0, or from unstable => upcoming Debian 2.2 for glibc
> 2.1) from your nearest Debian mirror (see http://www.debian.org/)
> and converting them to RPM packages via alien. You can also unpack
> Debian packages manually (they are ar archives, with two tarballs
> inside, one with the files, the other can be ignored). The packages
> are in the devel sections of each distribution.
I can attest that the latest unstable Debian packages work fine on Red
Hat 6.0. You'll just need alien which can be found at
http://kitenet.net/programs/alien/
There's a nice work-in-progress document by Daniel Barlow about using
CMUCL on Linux at
http://www.telent.net/lisp/howto.html
To prove that Lisp on Linux is alive and well the document, dated
1999-06-01, mentions problems with glibc 2.1 which no longer exist.
Hannu Rummukainen
····@acm.org (Pierre R. Mai) writes:
Just to follow up on myself:
[ List of CL implementations on Linux elided]
> On the whole, the situation for CL on Linux/Intel seems better than
> ever before, with three major commercial implementations available
> now, and at least two open-source implementations available and going
> strong. And even non-Intel Linux plattforms seem to be getting more
> implementations (I think CLISP has supported some of them for some
> time) nowadays. And that isn't counting no-longer supported
> implementations like GCL & Co.
Just remembered that I had forgotten to include the newest entry into
the free Linux implementations arena, namely Poplog! (silly me!)
Poplog is an implementation of Pop11, ML, Common Lisp and Prolog on
top of a common substrate, now freely available at:
ftp://ftp.cs.bham.ac.uk/pub/dist/poplog/freepoplog.html
So this brings the count of current CL implementations for Linux to 3
commercial and 3 open source implementations, plus a couple of other
less current implementations. There are few languages that can boast
more implementations on Linux(Intel), IMNSHO...
Regs, Pierre.
--
Pierre Mai <····@acm.org> PGP and GPG keys at your nearest Keyserver
"One smaller motivation which, in part, stems from altruism is Microsoft-
bashing." [Microsoft memo, see http://www.opensource.org/halloween1.html]
On 07 Aug 1999 14:01:04 +0200, Pierre R. Mai <····@acm.org> wrote:
>"David McClain" <········@azstarnet.com> writes:
>
>In recent times the frequency of "Is XXX Dead?" threads in c.l.l have
>increased to the point that theories about general world weariness at
>the end of a century/millenium seem probable again. Maybe this should
>be called the "Human Millenium Bug", or shorter "HUMBUG". Or maybe it
>is just the solar eclipse, and there has been too much reading of
>Nostradamus... ;)
Would anyone object if we turn something like this into a periodic
posting? It would help out the ignorant and the enlightened :-)
--
Kenneth P. Turvey <·······@SprocketShop.com>
----------------- http://www.tranquility.net/~kturvey
There are two major products that come out of Berkeley: LSD and UNIX.
We don't believe this to be a coincidence.
-- Jeremy S. Anderson
From: ········@my-deja.com
Subject: Re: Is the King Dead? (was Re: Is Lisp Dead on Linux? (Far from it!))
Date:
Message-ID: <7oicpr$krf$1@nnrp1.deja.com>
A touch of the fin de siecle blues perhaps? Probably more constructive
than throwing yourself off the nearest promotory ;-)
Fortunately, fewer people may top themselves this time. :-|
My personal experience of cl under Slackware linux running on a P75
5x86 with egcs 1.1.2 and glibc 2.1 is:
The debian CMUCL 2.4.13 runs well (installed using alien and un-taring
binaries). This will not bootstrap itself using CMUCL 2.4.13 source and
I don't know why.
The CLISP 1999-01 runs fine (installed using source and compiling).
Some problems compiling with the 1999-07 distribution.
I also got GCL-2.2 to compile (with a bit of 'adjusting').
Having just installed gcc-2.95, I will recompile in the next few days.
But on a P75 it will take a few days.
Best Regards,
:-) will
ps: You should never forget walk ;-)
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.
From: Bruno Haible
Subject: New CLISP release (was Re: Is the King Dead? (was Re: Is Lisp Dead on Linux? (Far from it!)))
Date:
Message-ID: <7onl2a$ga3$1@news.u-bordeaux.fr>
Pierre R. Mai <····@acm.org> wrote:
>
> I can't speak for CLISP, but since I've seen the main author of CLISP
> active here a couple of days ago, and he made no indication at all
> that CLISP is dead (with versions from 1999 available), I don't think
> that CLISP is dead either.
A new version of CLISP has been released two weeks ago. Highlights:
* Unicode support.
* Interface to PostgreSQL.
* Implements the Gray proposal for user-defined streams.
* Better ANSI CL compliance.
On ftp://clisp.cons.org/pub/lisp/clisp/
and ftp://metalab.unc.edu/pub/Linux/devel/lang/lisp/
with binaries for
i586-linux-libc6
i586-linux-libc5
alpha-linux
sparc-linux
m68k-linux
> And even non-Intel Linux plattforms seem to be getting more
> implementations (I think CLISP has supported some of them for some
> time) nowadays.
sparc-linux and m68k-linux are pretty recent ports. The first i386-linux
port of CLISP, however, dates back to Feb 13, 1992 (Linux 0.12).
Bruno http://clisp.cons.org/
"David McClain" <········@azstarnet.com> writes:
> Well doggone anyway... I wanted to get a Lisp up and running on my new RH 6
> Linux box and both CMUCL and CLISP bomb out before I can even say hello...
[]
> Libc. CLISP, OTOH bombs with a bad file descriptor.
IIRC, that sounds like the same problem I had, and I got around it by
building it from source. What version of glibc do you have? The
CLISP binaries for redhat are (were?) linked with a development
version.
> It appears that the
> CMUCL crowd went on to CMU Dylan, which is also dead from the sounds of it,
> and Harlequin is also disappearing... So, doesn't anyone use Lisp on Linux?
I do. The Lisp situation on Linux is not as rosy as I'd like it, tho.
--
Tom Breton, http://world.std.com/~tob
Ugh-free Spelling (no "gh") http://world.std.com/~tob/ugh-free.html
Tom Breton <···@world.std.com> writes:
> I do. The Lisp situation on Linux is not as rosy as I'd like it, tho.
Could you elaborate on that? It seems to me that CL's situation on
Linux is really fine, nowadays. What do you find missing to make the
situation rosy?
Regs, Pierre.
--
Pierre Mai <····@acm.org> PGP and GPG keys at your nearest Keyserver
"One smaller motivation which, in part, stems from altruism is Microsoft-
bashing." [Microsoft memo, see http://www.opensource.org/halloween1.html]
From: Lars Bj�nnes
Subject: Re: Common Lisp is doing great on Linux (was Re: Is Lisp Dead on Linux?)
Date:
Message-ID: <m3d7wxs50u.fsf@enterprise.gdpm.no>
·····@sip.medizin.uni-ulm.de (kp gores) writes:
> In article <·················@orion.dent.isdn.cs.tu-berlin.de>,
> ····@acm.org (Pierre R. Mai) wrote:
>
> >Could you elaborate on that? It seems to me that CL's situation on
> >Linux is really fine, nowadays. What do you find missing to make the
> >situation rosy?
>
> support for macintosh, i.e. powerpc.
I think Franz has ported ACL to LinuxPPC/MkLinux. Check their website.
--
Lars
·····@sip.medizin.uni-ulm.de (kp gores) writes:
> In article <·················@orion.dent.isdn.cs.tu-berlin.de>,
> ····@acm.org (Pierre R. Mai) wrote:
>
> >Tom Breton <···@world.std.com> writes:
> >
> >> I do. The Lisp situation on Linux is not as rosy as I'd like it, tho.
> >
> >Could you elaborate on that? It seems to me that CL's situation on
> >Linux is really fine, nowadays. What do you find missing to make the
> >situation rosy?
>
> support for macintosh, i.e. powerpc.
Yes, that's a real problem (and not only for PPC, but for all
non-intel Linux ports, like Alpha, (Ultra)Sparc, etc.). OTOH Lisp's
situation in this regard doesn't seem worse than that of most other
non-free program and/or complex programming languages.
But support for PPC does seem to be getting better: It seems Franz has
ported ACL to LinuxPPC. And I think that CLISP should either run
there already, or should be easy to port. Eclipse probably would be
easy to port as well. CMU CL would be much harder to port though.
I don't know about Poplog, but it sounds as if it might be ported with
reasonable effort.
Regs, Pierre.
--
Pierre Mai <····@acm.org> PGP and GPG keys at your nearest Keyserver
"One smaller motivation which, in part, stems from altruism is Microsoft-
bashing." [Microsoft memo, see http://www.opensource.org/halloween1.html]
····@acm.org (Pierre R. Mai) writes:
> Yes, that's a real problem (and not only for PPC, but for all
> non-intel Linux ports, like Alpha, (Ultra)Sparc, etc.). OTOH Lisp's
> situation in this regard doesn't seem worse than that of most other
> non-free program and/or complex programming languages.
CMUCL runs on the Alpha architecture. Whether the combination
Alpha/Linux does, I should check. Anyway, porting to Linux/Alpha
shouldn't be too difficult.
> But support for PPC does seem to be getting better: It seems Franz has
> ported ACL to LinuxPPC. And I think that CLISP should either run
> there already, or should be easy to port. Eclipse probably would be
> easy to port as well. CMU CL would be much harder to port though.
CMUCL requires a backend for PPC (AIX and/or LINUX should be
targeted). Any taker? :) (Don't look at me! I just threw the stone :) )
This is much more complex.
Cheers
--
Marco Antoniotti ===========================================
PARADES, Via San Pantaleo 66, I-00186 Rome, ITALY
tel. +39 - 06 68 10 03 17, fax. +39 - 06 68 80 79 26
http://www.parades.rm.cnr.it/~marcoxa
Marco Antoniotti <·······@copernico.parades.rm.cnr.it> writes:
> ····@acm.org (Pierre R. Mai) writes:
>
> > Yes, that's a real problem (and not only for PPC, but for all
> > non-intel Linux ports, like Alpha, (Ultra)Sparc, etc.). OTOH Lisp's
> > situation in this regard doesn't seem worse than that of most other
> > non-free program and/or complex programming languages.
>
> CMUCL runs on the Alpha architecture. Whether the combination
> Alpha/Linux does, I should check. Anyway, porting to Linux/Alpha
In a recent posting, someone pointed out to me that neither did
Linux/Alpha work natively at the time, nor did the DU/Alpha port work
via Linux's binary emulation support. OTOH I seem to remember that
someone posted recently on cmucl-imp about working on some port, but I
could be confusing this with some other port.
> shouldn't be too difficult.
Not too difficult, but not trivially easy either...
Regs, Pierre.
--
Pierre Mai <····@acm.org> PGP and GPG keys at your nearest Keyserver
"One smaller motivation which, in part, stems from altruism is Microsoft-
bashing." [Microsoft memo, see http://www.opensource.org/halloween1.html]
In article <······················@omikron.medizin.uni-ulm.de>, ·····@sip.medizin.uni-ulm.de (kp gores) wrote:
> In article <·················@orion.dent.isdn.cs.tu-berlin.de>,
> ····@acm.org (Pierre R. Mai) wrote:
>
> >Tom Breton <···@world.std.com> writes:
> >
> >> I do. The Lisp situation on Linux is not as rosy as I'd like it, tho.
> >
> >Could you elaborate on that? It seems to me that CL's situation on
> >Linux is really fine, nowadays. What do you find missing to make the
> >situation rosy?
>
> support for macintosh, i.e. powerpc.
> ciao
> kp
ACL 5.0.1 should be running on PowerPC Macs running Linux.
NASA has ported MCL to Linux on PPC.
From: Martin Cracauer
Subject: Re: Common Lisp is doing great on Linux (was Re: Is Lisp Dead on Linux?)
Date:
Message-ID: <7omosa$u3e$1@counter.bik-gmbh.de>
·····@sip.medizin.uni-ulm.de (kp gores) writes:
>In article <·················@orion.dent.isdn.cs.tu-berlin.de>,
>····@acm.org (Pierre R. Mai) wrote:
>>Tom Breton <···@world.std.com> writes:
>>
>>> I do. The Lisp situation on Linux is not as rosy as I'd like it, tho.
>>
>>Could you elaborate on that? It seems to me that CL's situation on
>>Linux is really fine, nowadays. What do you find missing to make the
>>situation rosy?
>support for macintosh, i.e. powerpc.
CMUCL once had a backend for the IBM PC/RT, which is a precursor for
the RS/6000, which is a precursor for the PowerPC.
I don't know how much backward compatiblity these chip generations
carry around, but it may be reasonably easy to get a port that is
working, although not generating efficient code.
Martin
--
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Martin Cracauer <········@bik-gmbh.de> http://www.bik-gmbh.de/~cracauer/
"Where do you want to do today?" Hard to tell running your calendar
program on a junk operating system, eh?
From: Hartmann Schaffer
Subject: Re: Common Lisp is doing great on Linux (was Re: Is Lisp Dead on Linux?)
Date:
Message-ID: <gIMr3.37239$5r2.72439@tor-nn1.netcom.ca>
In article <············@counter.bik-gmbh.de>,
········@bik-gmbh.de (Martin Cracauer) writes:
> ...
> CMUCL once had a backend for the IBM PC/RT, which is a precursor for
> the RS/6000, which is a precursor for the PowerPC.
>
> I don't know how much backward compatiblity these chip generations
> carry around, but it may be reasonably easy to get a port that is
> working, although not generating efficient code.
THere is hardly any similarity between the PC/RT chip and the other two
chips: the only similarity is that they are both RISC chips.
--
Hartmann Schaffer
It is better to fill your days with life than your life with days
From: Tom Breton
Subject: Re: Common Lisp is doing great on Linux (was Re: Is Lisp Dead on Linux?)
Date:
Message-ID: <m3oggg2uzi.fsf@world.std.com>
····@acm.org (Pierre R. Mai) writes:
> Tom Breton <···@world.std.com> writes:
>
> > I do. The Lisp situation on Linux is not as rosy as I'd like it, tho.
>
> Could you elaborate on that? It seems to me that CL's situation on
> Linux is really fine, nowadays. What do you find missing to make the
> situation rosy?
Perhaps we're comparing it to different things. You sound like you're
comparing it to an earlier situation that I know little of, whereas
I'm spoiled by the numerous C and Perl packages that built without
trouble. I've never fetched a Lisp package that didn't need a little
tinkering before it would run.
--
Tom Breton, http://world.std.com/~tob
Ugh-free Spelling (no "gh") http://world.std.com/~tob/ugh-free.html
Tom Breton <···@world.std.com> writes:
> ····@acm.org (Pierre R. Mai) writes:
>
> > Tom Breton <···@world.std.com> writes:
> >
> > > I do. The Lisp situation on Linux is not as rosy as I'd like it, tho.
> >
> > Could you elaborate on that? It seems to me that CL's situation on
> > Linux is really fine, nowadays. What do you find missing to make the
> > situation rosy?
>
> Perhaps we're comparing it to different things. You sound like you're
> comparing it to an earlier situation that I know little of, whereas
> I'm spoiled by the numerous C and Perl packages that built without
> trouble. I've never fetched a Lisp package that didn't need a little
> tinkering before it would run.
We indeed seem to be talking about different things. Unless you claim
that the Lisp packages you fetched worked better under Lisps on
platforms other than Linux, I'd say though that your complaint isn't
really about the situation of Lisp on Linux, but more about the
situation of Lisp as a whole (or certain implementations) and/or about
the quality of Lisp packages. Agreed?
I have to say that using CMU CL I've had little problems with the
packages I've come across, but YMMV. Have you submitted
patches/advice to the providers of the Lisp packages you had problems
with (when the package maintainers where still active/reachable)?
It's sometimes difficult to support implementations you don't usually
use, especially if there is little feedback.
Regs, Pierre.
--
Pierre Mai <····@acm.org> PGP and GPG keys at your nearest Keyserver
"One smaller motivation which, in part, stems from altruism is Microsoft-
bashing." [Microsoft memo, see http://www.opensource.org/halloween1.html]
From: Tom Breton
Subject: Re: Common Lisp is doing great on Linux (was Re: Is Lisp Dead on Linux?)
Date:
Message-ID: <m3u2q7hzmy.fsf@world.std.com>
····@acm.org (Pierre R. Mai) writes:
> platforms other than Linux, I'd say though that your complaint isn't
> really about the situation of Lisp on Linux, but more about the
> situation of Lisp as a whole (or certain implementations) and/or about
> the quality of Lisp packages. Agreed?
Sure.
> I have to say that using CMU CL I've had little problems with the
> packages I've come across, but YMMV. Have you submitted
> patches/advice to the providers of the Lisp packages you had problems
> with (when the package maintainers where still active/reachable)?
Yes, where reasonable. For instance, I submitted my fix to make FUF
run under CLISP. I don't think the maintainer put it up, since there
hasn't been a new version for some time, but it's available thru my
web page.
--
Tom Breton, http://world.std.com/~tob
Ugh-free Spelling (no "gh") http://world.std.com/~tob/ugh-free.html
"David McClain" <········@azstarnet.com> writes:
>Well doggone anyway... I wanted to get a Lisp up and running on my new RH 6
>Linux box and both CMUCL and CLISP bomb out before I can even say hello...
>After pouring through the archives on CMUCL it appears that I have to wait
>for a new build (if one ever appears) because of fundamental changes to the
>Libc. CLISP, OTOH bombs with a bad file descriptor. It appears that the
>CMUCL crowd went on to CMU Dylan, which is also dead from the sounds of it,
>and Harlequin is also disappearing... So, doesn't anyone use Lisp on Linux?
All conspiration theory aside, here is why the FreeBSD port of CMUCL
appears to be in better maintainance than Linux:
1) We just don't have a person that builds a Linux binary on every
occasion. It's a simple matter of volunteers. Yes we know it isn't
easy to build CMUCL and a new build infrastructure would help. We
need volunteers for that also.
2) The two developers that made the x86 going and the person that runs
the cons.org infrastructure (and does the integration of CMUCL into
FreeBSD releases) run FreeBSD. When serious changes are made to
CMUCL, the developer hopefully puts up a binary in addition to his
source patches and in most cases this is a FreeBSD binary.
3) FreeBSD with its single-distribution/single-ftp-server
infrastructure makes it easier to spot the newest FreeBSD version.
In FreeBSD, you just get the newest ports/lang/cmucl and can be
sure to have a reasonable new CMUCL. For Linux, you have to look up
the newest stuff and you can't tell which distribution has it. For
now, our Linux release builder provides his stuff as Debian
.deb-files, where his old rpms still float around the archives in
great number. If you're a Linux user, you have to learn how to
handle this situation.
4) The Linux port is in some regards in even better shape: The *.deb
files do a rebuild from source if you want it. Also the FreeBSD
version of CMUCL is still aout. It runs on FreeBSD/ELF (3.x and
4.x), but you have to use aout binaries to use alien objects.
5) If you're trying CMUCL on Linux and fall over the libc symbol issue
(_setfpcw or so), you face exactly the reason why *I* run a system
where libc, kernel and compiler are maintained by *one* group of
persons, a group where *one* way of setting the initial value of
the FPU status word is followed and where programs without floating
point instructions don't use the FPU (and therefore improve task
switch time). If you're running Linux, you've got to learn how to
handle this situation.
Martin
--
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Martin Cracauer <········@bik-gmbh.de> http://www.bik-gmbh.de/~cracauer/
"Where do you want to do today?" Hard to tell running your calendar
program on a junk operating system, eh?
Martin Cracauer <········@bik-gmbh.de> wrote:
>
>5) If you're trying CMUCL on Linux and fall over the libc symbol issue
> (_setfpcw or so), you face exactly the reason why *I* run a system
> where libc, kernel and compiler are maintained by *one* group of
> persons ...
The way GNU and Linux development is organized has nothing to do with
the __setfpucw issue. __setfpucw was an undocumented function. The "__"
prefix indicates that. Programs like CMUCL and CLISP should never have
used __setfpucw, but at a certain moment is was the most promising approach
for modifying FPU flags in a portable way.
Bruno
"David McClain" <········@azstarnet.com> writes:
> Well doggone anyway... I wanted to get a Lisp up and running on my
> new RH 6 Linux box and both CMUCL and CLISP bomb out before I can
> even say hello... After pouring through the archives on CMUCL it
> appears that I have to wait for a new build (if one ever appears)
> because of fundamental changes to the Libc. CLISP, OTOH bombs with a
> bad file descriptor. It appears that the CMUCL crowd went on to CMU
> Dylan, which is also dead from the sounds of it, and Harlequin is
> also disappearing... So, doesn't anyone use Lisp on Linux?
i use lisp on linux -- i am even using the redhat 6.0 flavor. i've
gotten allegro common-lisp trial edition, cmucl and clisp all working.
ACL for glibc-2.0 works just fine with the newer libc. there's a
cmucl specifically for glibc-2.1 on the ftp.cons.org site. i think
clisp should run after a recompile. get the source rpm and --rebuild
it.
hope this helps
--
johan kullstam
CLISP certainly works straight off on SuSE Linux (at least v5.0). I am
switching to Debian when the next release appears - partly because the
Linux CMUCL is maintained by a Debian user. I've just received a CD in
the post from Franz with a Linux Lisp which is in a RedHat format (I
haven't installed it yet because of the impending switch to Debian...).
Also Poplog has become freely available over the last month or so and
incldues Lisp (although of all those I believe CMUCL and Franz will give
the best performance if that's important - the free Franz Lisp comes
with licence restrictions however). Both CMUCL and Dylan are maintained
by people outside CMU these days, I think.
So there's a lot of implementations (have you checked out
http://www.elwoodcorp.com/alu/table/contents.htm for more info?) around
which *should* work. If you have problems, try containing the
maintainer of whatever program you are using - they will appreciate any
bug reports (the problem with libc is because there's been a major
change in Linux recently - that's what happens when improvements break
existing binaries, and if it wasn't for open source you'd be in a *real*
mess :-)
Andrew
In article <·············@corp.supernews.com>,
"David McClain" <········@azstarnet.com> wrote:
> Well doggone anyway... I wanted to get a Lisp up and running on my new
RH 6
> Linux box and both CMUCL and CLISP bomb out before I can even say
hello...
> After pouring through the archives on CMUCL it appears that I have to
wait
> for a new build (if one ever appears) because of fundamental changes
to the
> Libc. CLISP, OTOH bombs with a bad file descriptor. It appears that
the
> CMUCL crowd went on to CMU Dylan, which is also dead from the sounds
of it,
> and Harlequin is also disappearing... So, doesn't anyone use Lisp on
Linux?
>
> D.McClain
>
>
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.