Hi again. My real goal is to get CLM/CM working on windows XP. If any
one has gotten it to work I would love to know how and then the rest of
this post is irrelevant.
I gave up trying to do it through allegro/lispbox directly. It seems
there were lots of problems with packages and/or pathnames needing \
and getting / instead. Since I am a noob I gave up on that pretty
fast.
My next try was with cygwin/clisp. I got to the point where CM (common
music) loaded up with only one error (a problem with something called
GNU-PLOT) but CLM still wouldn't load up without multiple errors. Also
most of the other packages that work with CM wouldn't load either. The
only one that worked was CFFI, whatever that is.
So I figured I would try using CMUCL instead of CLISP since it is the
version of lisp that allows the most CM
packages/applications/extensions to be used.
Which brings me to my question. CMUCL needs a working version of CMUCL
to create a working version of CMUCL? Am I right here? I can't be
right. That is INSANE. I tried something called cross-compilation
using CLISP which got halfway through and then failed. What is going
on here? I mean... there has got to be a simpler way right?
P.S. I don't want to start a flame war here and I don't want to
discourage people from answering my question but are Linux/Unix people
sadistic or lazy? I mean... why do I have to jump through so many
hoops to do something that a single install executable on XP would do
with one click? There is something here I am just not getting, I
suppose. Maybe it is purposefully hard: to weed the "unintelligent"
out and make everyone feel elite and smart?
I am sorry I am ranting here but I just want to learn lisp on a noobie
level and start using CLM, not become the linux install & lisp package
master of the universe.
On Jan 25, 2:39 pm, "landspeedrecord" <···············@gmail.com>
wrote:
> Which brings me to my question. CMUCL needs a working version of CMUCL
> to create a working version of CMUCL? Am I right here? I can't be
> right. That is INSANE.
Why? How easy do you think it would be to compile, say, gcc, without a
working gcc? Or to build Windows without a working version of Windows
(do you think they cross-compile it on Solaris, or what?).
The more interesting question is: why are you trying to compile CMUCL?
Either there are binaries of it which run on Windows already, or it's a
bit unlikely that you will be able to create them.
On Jan 25, 2:39 pm, "landspeedrecord" <···············@gmail.com>
wrote:
> Hi again. My real goal is to get CLM/CM working on windows XP. If any
> one has gotten it to work I would love to know how and then the rest of
> this post is irrelevant.
>
> I gave up trying to do it through allegro/lispbox directly. It seems
> there were lots of problems with packages and/or pathnames needing \
> and getting / instead. Since I am a noob I gave up on that pretty
> fast.
>
> My next try was with cygwin/clisp. I got to the point where CM (common
> music) loaded up with only one error (a problem with something called
> GNU-PLOT) but CLM still wouldn't load up without multiple errors. Also
> most of the other packages that work with CM wouldn't load either. The
> only one that worked was CFFI, whatever that is.
GNUPLOT is a separate program for graph-plotting it has been ported to
Windows but would have to be installed separately.
Cliki (http://www.cliki.net/Common%20Lisp%20Music) indicates that
Common Lisp music only works with Allegro CL or Clisp on Windows. I
have no idea how to install it though. I assume if you're asking here
then the manual is no help.
> So I figured I would try using CMUCL instead of CLISP since it is the
> version of lisp that allows the most CM
> packages/applications/extensions to be used.
>
> Which brings me to my question. CMUCL needs a working version of CMUCL
> to create a working version of CMUCL? Am I right here? I can't be
> right. That is INSANE. I tried something called cross-compilation
> using CLISP which got halfway through and then failed. What is going
> on here? I mean... there has got to be a simpler way right?
It's generally the same with C compilers since they're written in C,
just as
Lisp compilers are written in lisp. That's why both are often
distributed as binaries.
The reason you haven't found one for CMUCL is that it doesn't support
Windows.
<grumpy ranting snipped>
What I can tell you is that the Lisps that work on windows are SBCL,
Clisp, ECL and GCL.
All provide executable downloads from their websites that require no
building from source.
On Jan 25, 3:47 pm, "Rob Thorpe" <·······@realworldtech.com> wrote:
> What I can tell you is that the Lisps that work on windows are SBCL,
> Clisp, ECL and GCL.
And LispWorks and ACL which both provide slightly restricted free
versions.
--tim
First off let me say thanks for the fast replys!
My reply to Tim:
Tim said: "How easy do you think it would be to compile, say, gcc,
without a
working gcc? Or to build Windows without a working version of Windows.
The more interesting question is: why are you trying to compile CMUCL?"
I am not trying to compile it on windows. I am using Cygwin (which I
assume is a linux emulator of some sort) and I couldn't find any binary
versions of it so I got the tar ball which only had source. If there
is a linux/unix binary that works under cygwin then great I will use
that but I didn't see one.
The windows comment I think is a bad analogy as windows is an operating
system. CMUCL is not. I guess my grumbly rant's point is that...
let's assume I use Linux and not an emulator here... what is so wrong
with an installation executable? Are linux people allergic to it? Why
use cvs and tar and builds etc? I apologize if I am making a faulty
assumption... perhaps the reason for my difficulties is that I am using
an emulator of linux and not linux itself.
My reply to Ken:
Thanks for the mailing list info. I signed up for it yesterday though,
but I do appreciate you passing it along. I would have posted there
today but the whole "CMUCL needs CMUCL" thing seemed better placed in a
more general forum like this one. I will try not to spend the money
too fast once I cash your check. Is it okay if I add an extra zero?
My reply to Rob:
Thanks for the info on GNU-PLOT but if I am under cygwin shouldn't it
work right off? Or am I missing something? I certainly may be since I
have no idea what I am doing. As for the compiling thing, see my reply
to Tim above. My problem is that CLM is fussy and seems not to want to
run under windows directly so I am trying to sneak in the back door by
running it under a linux emulator (CYGWIN) which is turning out to be
just as difficult and unsucessful. Plus I am stubborn and now that I
can't get CMUCL working I want to just because.
On Jan 25, 10:47 am, "Rob Thorpe" <·······@realworldtech.com> wrote:
> On Jan 25, 2:39 pm, "landspeedrecord" <···············@gmail.com>
> wrote:
>
> > Hi again. My real goal is to get CLM/CM working on windows XP. If any
> > one has gotten it to work I would love to know how and then the rest of
> > this post is irrelevant.
>
> > I gave up trying to do it through allegro/lispbox directly. It seems
> > there were lots of problems with packages and/or pathnames needing \
> > and getting / instead. Since I am a noob I gave up on that pretty
> > fast.
>
> > My next try was with cygwin/clisp. I got to the point where CM (common
> > music) loaded up with only one error (a problem with something called
> > GNU-PLOT) but CLM still wouldn't load up without multiple errors. Also
> > most of the other packages that work with CM wouldn't load either. The
> > only one that worked was CFFI, whatever that is.GNUPLOT is a separate program for graph-plotting it has been ported to
> Windows but would have to be installed separately.
> Cliki (http://www.cliki.net/Common%20Lisp%20Music) indicates that
> Common Lisp music only works with Allegro CL or Clisp on Windows. I
> have no idea how to install it though. I assume if you're asking here
> then the manual is no help.
>
> > So I figured I would try using CMUCL instead of CLISP since it is the
> > version of lisp that allows the most CM
> > packages/applications/extensions to be used.
>
> > Which brings me to my question. CMUCL needs a working version of CMUCL
> > to create a working version of CMUCL? Am I right here? I can't be
> > right. That is INSANE. I tried something called cross-compilation
> > using CLISP which got halfway through and then failed. What is going
> > on here? I mean... there has got to be a simpler way right?It's generally the same with C compilers since they're written in C,
> just as
> Lisp compilers are written in lisp. That's why both are often
> distributed as binaries.
> The reason you haven't found one for CMUCL is that it doesn't support
> Windows.
>
> <grumpy ranting snipped>
>
> What I can tell you is that the Lisps that work on windows are SBCL,
> Clisp, ECL and GCL.
> All provide executable downloads from their websites that require no
> building from source.
On Jan 25, 4:20 pm, "landspeedrecord" <···············@gmail.com>
wrote:
> I am not trying to compile it on windows. I am using Cygwin (which I
> assume is a linux emulator of some sort) and I couldn't find any binary
> versions of it so I got the tar ball which only had source. If there
> is a linux/unix binary that works under cygwin then great I will use
> that but I didn't see one.
As I said, if there isn't a binary for your platform, then there's
basically no chance you will be able to make one.
> The windows comment I think is a bad analogy as windows is an operating
> system. CMUCL is not.
No, but it is a large, complex system which has significant
dependencies on details of the OS at a level which cygwin probably
can't hide.
But even if it would build on cygwin, it's perfectly reasonable that
you would need a preexisting installation: that is the common case for
compilers (there are exceptions of course: the first time I built gcc I
did it starting from the pcc descendant that shipped with SunOS 4, and
it may still be possible to compile it with a non-gcc compiler I
suppose).
> I guess my grumbly rant's point is that...
> let's assume I use Linux and not an emulator here... what is so wrong
> with an installation executable? Are linux people allergic to it? Why
> use cvs and tar and builds etc? I apologize if I am making a faulty
> assumption... perhaps the reason for my difficulties is that I am using
> an emulator of linux and not linux itself.
Nothing is wrong with that (well, on Unixoid systems you'd usually ship
some kind of package file such as an RPM or deb or soemthing, not an
executable). I'm not a current CMUCL user but I'd not consider
building it from source if I was (in fact, despite being one of the
very earliest users of CMUCL on non-Mach systems, I have never built it
from source). I think it typically has been distributed as a tar file
which just contains the binaries, libs etc, which is pretty close to an
installer really (I bet some desktop environments treat a tar file as
something you can click on and it will ask you where you want to read
it back to). I don't know if CMUCL ships as (say) an RPM, but I don't
see why it should not. If it doesn't it's probably because there is
finite resource available and people are busy doing other stuff.
--tim
From: Don Geddis
Subject: Re: Is CMUCL insane? P.S. Me vs. CLM (Common Lisp Music)
Date:
Message-ID: <877ivadiho.fsf@geddis.org>
Tim said:
> How easy do you think it would be to compile, say, gcc, without a working
> gcc? Or to build Windows without a working version of Windows. The more
> interesting question is: why are you trying to compile CMUCL?"
"landspeedrecord" <···············@gmail.com> wrote on 25 Jan 2007 08:2:
> I am not trying to compile it on windows. I am using Cygwin (which I
> assume is a linux emulator of some sort) and I couldn't find any binary
> versions of it
That's probably a clue that your goal will fail.
CMUCL's supported platforms are listed here:
http://www.cons.org/cmucl/platforms.html
> so I got the tar ball which only had source.
> Which brings me to my question. CMUCL needs a working version of CMUCL
> to create a working version of CMUCL? Am I right here? I can't be
> right. That is INSANE.
Now you're just being silly.
It turns out that compiling CMUCL from source is indeed difficult, and this
was probably the primary motivation for breaking off the SBCL project. SBCL
is essentially the same as CMUCL, but much easier to build (from any ANSI CL,
not just from CMUCL).
But get real. You sound shocked that building a compiler from source requires
a working version of that compiler. Have you ever built ANY compiler, for
ANY language, from source? How do you think that ANYBODY does it?
> The windows comment I think is a bad analogy as windows is an operating
> system. CMUCL is not.
So, then, answer Tim's initial question about GCC. Surely that is a
reasonable analogy. How would you recompile GCC on Cygwin, if you didn't
already have a working GCC? Have you ever tried anything like that?
> I guess my grumbly rant's point is that... let's assume I use Linux and
> not an emulator here... what is so wrong with an installation executable?
> Are linux people allergic to it? Why use cvs and tar and builds etc? I
> apologize if I am making a faulty assumption... perhaps the reason for my
> difficulties is that I am using an emulator of linux and not linux itself.
Before you complain so much, did you even take a moment to see what
installable executables were available? Here are a huge number of
pre-compiled binaries for CMUCL:
http://common-lisp.net/project/cmucl/downloads/binaries/
As well as, on the main download page
http://www.cons.org/cmucl/download.html
under "Alternative binary distributions", it mentions FreeBSD, Red Hat RPM,
Gentoo, and Debian. Under Debian linux, it's as easy as
root> apt-get install cmucl
just like any other Debian package.
Why did you EVER have the idea that "linux people" regularly use "cvs and
tar and builds" in order to install a working lisp?
> P.S. I don't want to start a flame war here and I don't want to
> discourage people from answering my question but are Linux/Unix people
> sadistic or lazy? I mean... why do I have to jump through so many
> hoops to do something that a single install executable on XP would do
> with one click? There is something here I am just not getting, I
> suppose. Maybe it is purposefully hard: to weed the "unintelligent"
> out and make everyone feel elite and smart?
Rather than being so rude, possibly the problem is simply that you don't
understand linux at all, rather than anything specific to lisp.
Just as XP has standard ways of installing new software packages, so too does
linux. XP tends to use GUIs, and linux tends to use command lines, but they're
both fairly straightforward once you get the hang of it.
I appreciate the trouble you went through, but it really seems like
1. most of your trouble was in trying to use cygwin (rather than some more
popular linux)
2. you were very insulting to the lisp (esp. cmucl) community, when in fact
most of what you demanded is ALREADY available, in exactly the form you
demanded it, and could have been found with some minor searching effort.
In any case, good luck with your lisp efforts under cygwin (or XP?), but
you shouldn't generalize your experience to what the typical linux person
encounters.
Hardly anybody (users) builds compilers from scratch.
-- Don
_______________________________________________________________________________
Don Geddis http://don.geddis.org/ ···@geddis.org
Mediocrity: It takes a lot less time and most people won't notice the
difference until it's too late. -- Despair.com
Thanks for reply Madhu.
Sorry that I didn't explain that CYGWIN is a linux/unix emulator. At
least I think that is what it is. I assumed everyone on this list
would know what it was, which I guess is silly of me. As for how I was
able to start the cross compile with CLISP... Maybe it wasn't CMUCL. I
had tried to cross compile SBCL too. Maybe that was the one that I was
thinking of. I am not sure anymore, I have tried so many different
combinations of Lisp interpreters and CM & CLM packages I can't even
keep track of them all at this point.
On Jan 25, 5:59 pm, "landspeedrecord" <···············@gmail.com>
wrote:
> Sorry that I didn't explain that CYGWIN is a linux/unix emulator. At
> least I think that is what it is. I assumed everyone on this list
> would know what it was, which I guess is silly of me.
I think most people here know what Cygwin is. And it is not really an
emulator, it is a "Linux-like environment" as they say on their
webpage. It is similar enough to Linux/Unix to compile some of the
more popular Unix programs, but it is /not/ similar enough to compile
something as complicated as CMUCL without modifications. You'll have
more luck with SBCL - they have Windows binaries available, albeit stll
in a somewhat experimental state.
Cheers,
Edi.
From: Richard M Kreuter
Subject: Re: Is CMUCL insane? P.S. Me vs. CLM (Common Lisp Music)
Date:
Message-ID: <877ivbrp3w.fsf@progn.net>
"landspeedrecord" <···············@gmail.com> writes:
> Sorry that I didn't explain that CYGWIN is a linux/unix emulator.
It's important to note that Cygwin is a library that emulates an API;
it's not an emulator in the sense of a system for letting you run
unmodified Unix binaries.
Unfortunately, a Unix-like API is not by itself sufficient for
portably implementing very much, without a fair amount of additional
elbow-grease; you'll find a great deal of system dependent code in
most non-trivial Unix programs (the system dependent code might be
packaged as a library, but it's still there).
In the present case, CMUCL, like some other programs, is fairly
sensitive to some system dependent things, e.g., the operating
system's facilities for memory mappings and signal handling.
Windows's facilities for these are different from Unix's; Cygwin,
which is merely a library, can't do anything to hide this fact from
applications. CMUCL could be implemented differently than it is, but
that's beside the point: the program is what it is, and nobody is
obliged to make it otherwise.
You might find that you'll reach your stated goal (CLM) faster if you
ask people familiar with CM for help getting CLM working with one of
the Lisps that already run on Windows.
--
RmK
Thanks Richard & Edi for you replys as well.
I guess my assumption that CYGWIN was a linux emulator is faulty and so
I have been barking up the wrong tree totally in thinking I could go
from source to binaries and get CM & CLM working properly.
So... 3 questions now then:
1) People are gently suggesting I take my quest for CML on XP off this
usenet group so feel free to ignore this one... but, is trying to run
CLM under CYGWIN a waste of time? I did get CM and CFFI working, keep
in mind.
2) is it a waste of time trying to get SBCL or CMUCL running under
CYGWIN? Seems like the answer is not only yes but it undoable without
lots of effort.
3) If all but the most trivial unix code have system dependent code
then what is CYGWIN good for? Feel free to ignore this question too
since I suppose I should be on a CYGWIN listserv asking about it at
this point.
On Jan 25, 1:01 pm, "landspeedrecord" <···············@gmail.com>
wrote:
[...]
> 2) is it a waste of time trying to get SBCL or CMUCL running under
> CYGWIN? Seems like the answer is not only yes but it undoable without
> lots of effort.
Of all the possible implementations of Common Lisp that are readily
available, CMUCL is one of the worst choices for working under Windows.
There are three commercial Common Lisps (Corman, Allegro and LispWorks)
that are available for Windows, and three free/OSS implementations as
well (CLISP, ECL and, experimentally, SBCL).
Do yourself a favor and try using *all* of them before considering
CMUCL.
Cheers,
Pillsy
From: Zach Beane
Subject: Re: Is CMUCL insane? P.S. Me vs. CLM (Common Lisp Music)
Date:
Message-ID: <m38xfqsyt9.fsf@unnamed.xach.com>
"landspeedrecord" <···············@gmail.com> writes:
> 2) is it a waste of time trying to get SBCL or CMUCL running under
> CYGWIN? Seems like the answer is not only yes but it undoable without
> lots of effort.
SBCL runs directly under Windows. It does not use Cygwin.
Zach
From: Ken Tilton
Subject: Re: Is CMUCL insane? P.S. Me vs. CLM (Common Lisp Music)
Date:
Message-ID: <Y78uh.1088$sJ3.36@newsfe10.lga>
landspeedrecord wrote:
> 1) People are gently suggesting I take my quest for CML on XP off this
> usenet group so feel free to ignore this one...
You talkin bout me? If so, no, I just think it obvious that if you are
looking for people who have CLM working you might look on a CLM list.
Too easy?
Meanwhile, I think you should continue to post here because we will be
able to explain Lisp errors and at some point I can likely help you,
having gotten CLM and CMN working on XP a coupla years ago. And if you
really get desperate I could ship you the version I had working, tho bit
rot likely has set in. Clearly, tho, it is better if you get the current
version working.
Your CMUCL thing has taken on a life of its own. This is classic. You
wanted to learn lisp and decided to do so in the context of CLM. Good
idea! You had problems with ACL and jumped to another Lisp. You should
have just solved the problems with ACL (by asking on the CLM list). But
you had decided the problem was with ACL, or that CMUCL would be easier.
So now your problem is with Cygwin, which is turning out not to be a
Unix. I predict your next post will be about getting Windows 95
installed so you can run a better Unix emulator, or maybe Knoppix so you
can run Linux so you can build CMUCL from source so you can use CLM to
do what you wanted to do: learn Lisp.
It happens.
kt
--
Well, I've wrestled with reality for 35 years, Doctor, and
I'm happy to state I finally won out over it.
-- Elwood P. Dowd
In this world, you must be oh so smart or oh so pleasant.
-- Elwood's Mom
On Jan 25, 8:09 pm, Ken Tilton <·········@gmail.com> wrote:
> landspeedrecord wrote:
>
> So now your problem is with Cygwin, which is turning out not to be a
> Unix. I predict your next post will be about getting Windows 95
> installed so you can run a better Unix emulator, or maybe Knoppix so you
> can run Linux so you can build CMUCL from source so you can use CLM to
> do what you wanted to do: learn Lisp.
>
This is one of those weird things where it's very hard to tell what is
happening. There are, I think, two common scenarios:
1. Person "wants to do x" but actually for various reasons either they
don't really want to do x, or they want to show that doing x is very
hard, or that x is badly designed, or something like that. So they
create endless hurdles for themselves and then complain loudly about
them. I think we've all seen this and we all know the solution
(tentacles, heads-on-stakes-being-pecked-by-ravens, that sort of
thing).
2. Person wants to do x, but manages to make maginificently bad
decisions when trying to do x. These are genuine mistakes however:
either they're unlucky or were badly advised or something.
I'm feeling nice today so I'll allow myself to believe that this might
be (2). I think then the two possibilities are: either get a proper
native Windows Lisp (of which there are a good number, including at
least two high-quality commercial ones which have slightly limited free
versions); or get a proper Linux/Unix and run a Linux/Unix Lisp on it.
If you want to do the latter, *without* destroying your existing
Windows installation then VMware is a very good solution, and there is
a perfectly adequate free version. You will need a decent amount of
memory (I found running a Solaris VMware instance with 4 zones on an XP
box a serious pain in 1G, but in 2G it was fine).
From: Ken Tilton
Subject: Re: Is CMUCL insane? P.S. Me vs. CLM (Common Lisp Music)
Date:
Message-ID: <6Cuuh.1261$sJ3.923@newsfe10.lga>
Tim Bradshaw wrote:
> On Jan 25, 8:09 pm, Ken Tilton <·········@gmail.com> wrote:
>
>>landspeedrecord wrote:
>>
>>So now your problem is with Cygwin, which is turning out not to be a
>>Unix. I predict your next post will be about getting Windows 95
>>installed so you can run a better Unix emulator, or maybe Knoppix so you
>>can run Linux so you can build CMUCL from source so you can use CLM to
>>do what you wanted to do: learn Lisp.
>>
>
>
> This is one of those weird things where it's very hard to tell what is
> happening. There are, I think, two common scenarios:
>
> 1. Person "wants to do x" but actually for various reasons either they
> don't really want to do x, or they want to show that doing x is very
> hard, or that x is badly designed, or something like that. So they
> create endless hurdles for themselves and then complain loudly about
> them. I think we've all seen this and we all know the solution
> (tentacles, heads-on-stakes-being-pecked-by-ravens, that sort of
> thing).
>
> 2. Person wants to do x, but manages to make maginificently bad
> decisions when trying to do x. These are genuine mistakes however:
> either they're unlucky or were badly advised or something.
>
> I'm feeling nice today so I'll allow myself to believe that this might
> be (2).
Well, the OP set no land speed records getting there, but s/he has now
landed on the shores of the CM mail list and is being attended to by
none other than Rick Taube Himself. So, yes, 2.
OTOH, it is a solid 2: they are committed now to making the end-run work
in which cygwin pretends to be Linux. But the list seems to be
encouraging this, so whadoiknow? (Except that that list is where they
would get the best help, in which case I wonder if they tried going back
to ACL or has somehow gotten stuck on the Linux under win32 thing--well,
there are signs they are getting close, and software likes to tease
people along on doomed approaches for weeks by almost working.
Well, look at the bright side: CLM is "free"! Hang in there LSR, and if
you want to try again on ACL I'll take a crack at it myself again to see
what I can see.
kzo
--
Well, I've wrestled with reality for 35 years, Doctor, and
I'm happy to state I finally won out over it.
-- Elwood P. Dowd
In this world, you must be oh so smart or oh so pleasant.
-- Elwood's Mom
From: Richard M Kreuter
Subject: Re: Is CMUCL insane? P.S. Me vs. CLM (Common Lisp Music)
Date:
Message-ID: <873b5ysrzo.fsf@progn.net>
"landspeedrecord" <···············@gmail.com> writes:
> 1) People are gently suggesting I take my quest for CML on XP off
> this usenet group so feel free to ignore this one... but, is trying
> to run CLM under CYGWIN a waste of time? I did get CM and CFFI
> working, keep in mind.
Cygwin is not particularly relevant here. The way to think about the
problem is how to get CLM working on some Lisp that runs on Windows.
The choices have been enumerated for you already in this thread.
> 2) is it a waste of time trying to get SBCL or CMUCL running under
> CYGWIN? Seems like the answer is not only yes but it undoable
> without lots of effort.
It would involve a lot of effort, probably not much less effort than
working on native ports of those implementations (i.e., ports of those
implementations that don't require a Unix-like environment).
However, SBCL already has a native Windows port, though I believe it
is still to be considered alpha-quality. But maybe it'll suffice for
you; see the SBCL project page for downloads.
> 3) If all but the most trivial unix code have system dependent code
> then what is CYGWIN good for? Feel free to ignore this question too
> since I suppose I should be on a CYGWIN listserv asking about it at
> this point.
All sorts of details vary from one Unix to another, including the
existence and values of limits for various system interfaces (e.g.,
maximum file sizes, maximum file name lengths), the availability and
names for details of optional features (networking, IIRC), and, of
course, the presence of bugs that need to be coded around. These
kinds of things are among what I was referring to as "system
dependencies", inasmuch as these are aspects of getting nontrivial
programs working on different Unix-like environments.
Cygwin provides an interface to the functionality normally found in a
Unix kernel. Along with a C library, a shell, a C compiler and a
collection of various other standard Unix tools, you can get an
environment on which you might be able to compile and run C programs
written for Unix. However, actually getting any particular Unix
program working in a Cygwin-based environment can require some effort,
the same as getting such programs working on any particular Unix
platform.
--
RmK
On Thu, 25 Jan 2007, landspeedrecord wrote:
> Thanks Richard & Edi for you replys as well.
>
> I guess my assumption that CYGWIN was a linux emulator is faulty and so
> I have been barking up the wrong tree totally in thinking I could go
> from source to binaries and get CM & CLM working properly.
>
> So... 3 questions now then:
> 1) People are gently suggesting I take my quest for CML on XP off this
> usenet group so feel free to ignore this one... but, is trying to run
Yes, you should ask on the cm/clm/cmn/snd mailing list, and also check out
the archives. I think there were some discussions about running clm/cm
in clisp recently (ie. not more than a couple of years ago).
> CLM under CYGWIN a waste of time? I did get CM and CFFI working, keep
> in mind.
>
No, it should work. I compiled SND under cygwin 2 or 3 years ago, and
it didn't complain about anything as far as I remember. And you can run
both CLM and CM in Snd, although in that case you would need to use scheme
and not common lisp. It might not be a bad alternative to start with
scheme though, in case you don't know lisp that well...?
Another alternative is CLISP. I've only skimmed the post in this thread,
but I couldn't see any signs of you trying CLISP. It should work too.
> 2) is it a waste of time trying to get SBCL or CMUCL running under
> CYGWIN? Seems like the answer is not only yes but it undoable without
> lots of effort.
>
Maybe, but both guile and gauche should work just fine in cygwin, and
clm/cm is supported by both of those via Snd.
In <························@v33g2000cwv.googlegroups.com>
landspeedrecord wrote:
> 3) If all but the most trivial unix code have system dependent code
> then what is CYGWIN good for? Feel free to ignore this question too
> since I suppose I should be on a CYGWIN listserv asking about it at
> this point.
>
CYGWIN will work for some C/C++ code.
I wouldn't take it too seriously, though.
Seriously, if plan to use Windows, get any implementation for Windows.
If you feel you are brave enough, take a plunge into Unix World with any
free Unix-like OS. I suggest PC-BSD (FreeBSD with easy package
management and graphical install - voted best easy instalation by some
magazine). It's free. Burn an iso disk. FreeBSD is not Linux, it's Unix.
You have CMUCL and SBCL there too.
JN
From: Ken Tilton
Subject: Re: Is CMUCL insane? P.S. Me vs. CLM (Common Lisp Music)
Date:
Message-ID: <dd4uh.7$M9.3@newsfe12.lga>
landspeedrecord wrote:
> Hi again. My real goal is to get CLM/CM working on windows XP. If any
> one has gotten it to work I would love to know how and then the rest of
> this post is irrelevant.
Pardon a meta-question, but if that is your goal, shouldn't you be
asking here? Very active, including Rick Taube himself:
(a) http://ccrma-mail.stanford.edu/mailman/listinfo/cmdist
(b) They are used to Noobs, it is a package meant for easy install in
that they try to make it Just Work automatically everywhere.
> P.S. I don't want to start a flame war here and I don't want to
> discourage people from answering my question but are Linux/Unix people
> sadistic or lazy?
Your check is in the mail. (In joke.)
kzo
--
Well, I've wrestled with reality for 35 years, Doctor, and
I'm happy to state I finally won out over it.
-- Elwood P. Dowd
In this world, you must be oh so smart or oh so pleasant.
-- Elwood's Mom
In <························@s48g2000cws.googlegroups.com>
landspeedrecord wrote:
used.
>
> Which brings me to my question. CMUCL needs a working version of
> CMUCL to create a working version of CMUCL? Am I right here? I can't
> be right. That is INSANE.
Languages like Lisp, Smalltalk and Forth have their own little "world."
Another well-known example is the Java VM. The little "world" of C is
the whole operating system. In fact, it's a huge "world."
To build something like the above, you need to hone in a certain OS,
with a minimal port, with a portable core, written in, say, C (why C?
because most OSs assume a certain architecture and have a C commpiler -
in another planet, we could have Lisp operating systems). Then, this
port starts compiling a language that compiles itself, in incremental
steps. I believe Allegro does that. Another example (non-Lisp) is Maple (
the CAS system), IIRC.
The ability of language of being capable of expressing aspects of itself
in the language itself is called "reification." For instance, Common
Lisp's expresses aspeccts of its obejct system in...Common Lisp.
Reification is an apsect of reflection. For instance, Perl had no object
system, but since the langauge allows reflection, the instructions of
the interpreted language are available to itself, so it can create
objects (which are functions with state) that it knows about.
Well, something like that...
Another approach, AFAIK, is simple carrying a binary to another OS,
provided they are similar enough in design (say, form NetBSD to OpenBSD).
From that point on you can build on that platform.
I think it's called bootstrapping, because you are raising yourself by
your shoestrings. It maybe that some Lisp system does not need that
bootstrappiing code - just the Lisp compiler. In which case, a lisp
system is needed to compile lisp. Just like you can take a newer gcc
version and use gcc to compile it (how else would you do it?)
Somebody correct me if I'm wrong.
JN
On Sat, 03 Feb 2007 00:46:17 +0000, joe nada wrote:
> In <························@s48g2000cws.googlegroups.com>
> landspeedrecord wrote:
> used.
>>
>> Which brings me to my question. CMUCL needs a working version of
>> CMUCL to create a working version of CMUCL? Am I right here? I can't
>> be right. That is INSANE.
>
>
> Languages like Lisp, Smalltalk and Forth have their own little "world."
> Another well-known example is the Java VM. The little "world" of C is
> the whole operating system. In fact, it's a huge "world."
This world is changing, and the changes are moving ever so slowly to the
mass-market desktop computing world. One of the fundamentals of the C
world is that each program runs on a single core and doesn't have any
parallelism. It's possible to retrofit parallelism onto C with libraries
and even (if you're evil) hacky little language extensions, but because
thinking about parallelism breaks most mundane programmers' brains we'll
have to switch to a language that allows the compiler to do the hard parts
for us. Common Lisp and Scheme are better than C, but Haskell beats them
both. (Prolog beats Haskell, but Prolog breaks brains even more readily
than source-level parallelism. ;))
> To build something like the above, you need to hone in a certain OS,
> with a minimal port, with a portable core, written in, say, C
This is how SBCL and CLisp do it. Depending on how much you think modules
modify the language, Perl might fit here as well.
> (why C?
> because most OSs assume a certain architecture and have a C commpiler -
> in another planet, we could have Lisp operating systems).
Already have done. See CONS, CADR, Symbolics, the TI Explorer, and the
Xerox Interlisp-D system, among others.
<http://fare.tunes.org/LispM.html>
Lisp machines tend to be Lisp all the way down to the CPU, such that the
microcode can be written in whatever proprietary Lisp dialect the machine
uses. (Lisp machines predate Common Lisp, and I don't think any of them
prior to the modern Scheme86 used Scheme.)
> Then, this
> port starts compiling a language that compiles itself, in incremental
> steps. I believe Allegro does that. Another example (non-Lisp) is Maple (
> the CAS system), IIRC.
> The ability of language of being capable of expressing aspects of itself
> in the language itself is called "reification." For instance, Common
> Lisp's expresses aspeccts of its obejct system in...Common Lisp.
> Reification is an apsect of reflection. For instance, Perl had no object
> system, but since the langauge allows reflection, the instructions of
> the interpreted language are available to itself, so it can create
> objects (which are functions with state) that it knows about.
> Well, something like that...
I don't think Perl's object system is as clean as all that, but it is
probably the closest thing to CLOS in the mainstream world.
> Another approach, AFAIK, is simple carrying a binary to another OS,
> provided they are similar enough in design (say, form NetBSD to OpenBSD).
> From that point on you can build on that platform.
Or you can build the binary on a completely different but more capable
system, like an x86 PC being used to build the compiler that will go on a
single-board computer built around an ARM CPU. This is called
cross-compilation and it is very common when building a foundation for an
entirely new computer system.
> I think it's called bootstrapping, because you are raising yourself by
> your shoestrings.
The expression is "pulling yourself up by your bootstraps", but you knew
that. ;)
> It maybe that some Lisp system does not need that
> bootstrappiing code - just the Lisp compiler. In which case, a lisp
> system is needed to compile lisp. Just like you can take a newer gcc
> version and use gcc to compile it (how else would you do it?) Somebody
> correct me if I'm wrong.
gcc can be built with any C compiler conforming to the 1990 ISO standard.
(Old versions could be built with pre-standard K&R compilers, but as of
3.4 that is no longer an option.) There are a lot of C compilers in the
world, but the 1990 standard is widely implemented and useful enough to
code gcc to. (That is only true of gcc's C frontend, as other frontends
(C++, Java, Objective-C, Fortran) may rely on gcc-specific code.
Presumable you could build a C-only gcc with the native C compiler, and
then use that to build the other frontends.)
>
> JN
--
My address happens to be com (dot) gmail (at) usenet (plus) chbarts,
wardsback and translated.
It's in my header if you need a spoiler.
----== Posted via Newsfeeds.Com - Unlimited-Unrestricted-Secure Usenet News==----
http://www.newsfeeds.com The #1 Newsgroup Service in the World! 120,000+ Newsgroups
----= East and West-Coast Server Farms - Total Privacy via Encryption =----
From: Zach Beane
Subject: Re: Is CMUCL insane? P.S. Me vs. CLM (Common Lisp Music)
Date:
Message-ID: <m3ejp7qqgc.fsf@unnamed.xach.com>
Chris Barts <··············@tznvy.pbz> writes:
> > To build something like the above, you need to hone in a certain OS,
> > with a minimal port, with a portable core, written in, say, C
>
> This is how SBCL and CLisp do it. Depending on how much you think modules
> modify the language, Perl might fit here as well.
The core of SBCL is written in Common Lisp. The GC and the runtime
(the bit that makes OS features easily accessible to Lisp) are written
in C. You can't bootstrap SBCL from a C compiler, but you can
bootstrap CLISP from a C compiler.
> > Reification is an apsect of reflection. For instance, Perl had no object
> > system, but since the langauge allows reflection, the instructions of
> > the interpreted language are available to itself, so it can create
> > objects (which are functions with state) that it knows about.
> > Well, something like that...
>
> I don't think Perl's object system is as clean as all that, but it is
> probably the closest thing to CLOS in the mainstream world.
Perl's object system consists mostly of a search ordering for looking
up methods given an object strongly associated with ("blessed" into) a
package, e.g. "What 'bar' does $foo->bar($baz) eventually call?". Oh
yeah, plus some magic for the SUPER package and AUTOLOAD. It is a
message-passing style. Everything else you get to roll yourself,
including object and slot representation (is it a hash reference, an
array reference, an inside-out object, or what?),
readers/writers/accessors, :before :after and :around methods,
dispatching on more than one argument, alternative method combination,
etc.
It's a far cry from CLOS.
Zach
On Sat, 03 Feb 2007 09:13:39 -0500, Zach Beane wrote:
> Chris Barts <··············@tznvy.pbz> writes:
>
>> > To build something like the above, you need to hone in a certain OS,
>> > with a minimal port, with a portable core, written in, say, C
>>
>> This is how SBCL and CLisp do it. Depending on how much you think modules
>> modify the language, Perl might fit here as well.
>
> The core of SBCL is written in Common Lisp. The GC and the runtime
> (the bit that makes OS features easily accessible to Lisp) are written
> in C. You can't bootstrap SBCL from a C compiler, but you can
> bootstrap CLISP from a C compiler.
Bah. Well, it's obviously been a very long time since I compiled SBCL.
>
>> > Reification is an apsect of reflection. For instance, Perl had no object
>> > system, but since the langauge allows reflection, the instructions of
>> > the interpreted language are available to itself, so it can create
>> > objects (which are functions with state) that it knows about.
>> > Well, something like that...
>>
>> I don't think Perl's object system is as clean as all that, but it is
>> probably the closest thing to CLOS in the mainstream world.
>
> Perl's object system consists mostly of a search ordering for looking
> up methods given an object strongly associated with ("blessed" into) a
> package, e.g. "What 'bar' does $foo->bar($baz) eventually call?". Oh
> yeah, plus some magic for the SUPER package and AUTOLOAD. It is a
> message-passing style. Everything else you get to roll yourself,
> including object and slot representation (is it a hash reference, an
> array reference, an inside-out object, or what?),
> readers/writers/accessors, :before :after and :around methods,
> dispatching on more than one argument, alternative method combination,
> etc.
>
> It's a far cry from CLOS.
It's closer than what either C++ or Java give you. I'm construing
'mainstream' pretty narrowly here, given the audience; some people might
say the only 'mainstream' language is COBOL.
--
My address happens to be com (dot) gmail (at) usenet (plus) chbarts,
wardsback and translated.
It's in my header if you need a spoiler.
----== Posted via Newsfeeds.Com - Unlimited-Unrestricted-Secure Usenet News==----
http://www.newsfeeds.com The #1 Newsgroup Service in the World! 120,000+ Newsgroups
----= East and West-Coast Server Farms - Total Privacy via Encryption =----
On 2007-02-03 14:13:39 +0000, Zach Beane <····@xach.com> said:
> The core of SBCL is written in Common Lisp. The GC and the runtime
> (the bit that makes OS features easily accessible to Lisp) are written
> in C. You can't bootstrap SBCL from a C compiler, but you can
> bootstrap CLISP from a C compiler.
I believe that you can bootstrap SBCL from CLISP however - I think that
was one of the reasons that it diverged from CMUCL. If this is true,
then by transitive closure, you actually can bootstrap SBCL from C :-)
--tim