matlisp-users

[Matlisp-users] Configuration failure on Linux/SBCL

[<dm...@ui...>]

Hello,

    After a long period of little lisp activity, I've just downloaded
the CVS version of matlisp via

    cvs -z3 -d:pserver:ano...@cv...:/cvsroot/matlisp\
        co matlisp

to try it with SBCL running on Gentoo Linux.  The INSTALL file begins

/----
| Requirements:
| =============
| Either CMUCL (version 18b), SBCL (version 0.8.5) or 
| Allegro CL (version 5.0 or later) is required.

No mention of SBCL there...

| * Allegro CL: Matlisp should compile on Linux, Solaris, Windows.
|               In particular, Matlisp relies on the foreign
|               function interface of Allegro CL.
| 
| * CMU CL: Matlisp should compile on Linux, Solaris.
|           In particular, Matlisp relies on the foreign function
|           interface of CMU CL and the built-in type 
|           kernel::complex-double-float.
| 
| * SBCL: Matlisp should compile on Linux and MacOS X
|         In particular, Matlisp relies on the foreign function
|         interface of SBCL.
\----

...but that last paragraph is encouraging.  The detailed instructions
include

/----
| 4. Configure the system.
|    o If you're using Allegro CL, specify --with-lisp=acl.
|    o If you're using CMU CL, specify --with-lisp=cmucl
|    o If you're using SBCL, specify --with-lisp=sbcl
|    o Give the name of the Lisp executable via --with-lisp-exec=<name>
|    o Run configure:
| 
|    ./configure --with-lisp=<lisp> --with-lisp-exec=<exec> --prefix=`pwd`
\----

so I tried

# ./configure --with-lisp=sbcl --with-lisp-exec=/usr/bin/sbcl --prefix=`pwd`

only to be told

checking build system type... i686-pc-linux-gnu
checking host system type... i686-pc-linux-gnu
checking target system type... i686-pc-linux-gnu
configure: error: Unknown Lisp system:  sbcl.  Supported systems are acl and cmucl

    Now, I know one plays with CVS versions at his own risk, and I'm
probably being dense to boot, but there's enough there to suggest
matlisp ought to work with SBCL that I thought it might be worth a
note.  Will someone please enlighten me about the status of SBCL
compatibility, or point out where I've gone wrong in the ./configure
line?  Thanks!

Michael

-- 
D. Michael McFarland
Department of Aerospace Engineering
University of Illinois at Urbana-Champaign
mailto:dm...@ui... https://netfiles.uiuc.edu/dmmcf/www/

Re: [Matlisp-users] Configuration failure on Linux/SBCL

[Robbie S. <rd...@me...>]

Hi,

I did the port of matlisp to sbcl.

Unfortunately, there were some problems with the patch (it got messed up 
by the sourceforge mailing list) so it wasn't completely applied to the 
cvs tree.  Parts of the patch were correctly applied, and others not.

I have a copy of my sources up at 
http://mercury.chem.pitt.edu/~rds/matlisp.sbcl.tar.gz

If you download these sources they should work out of the box. The only 
thing you might have to do is rebuild the configure script ("autoconf 
configure.in > configure").  After this the installation directions 
should work.

There is also in there an .asd system definition file, so if you want to 
use asdf to load matlisp you can just run configure as you did before, 
symlink matlisp.asd to the right place, and "(require :matlisp)" and it 
should compile and load everything.

Let me know if you have any problems.....

--Robbie

D. Michael McFarland wrote:
> Hello,
> 
>     After a long period of little lisp activity, I've just downloaded
> the CVS version of matlisp via
> 
>     cvs -z3 -d:pserver:ano...@cv...:/cvsroot/matlisp\
>         co matlisp
> 
> to try it with SBCL running on Gentoo Linux.  The INSTALL file begins
> 
> /----
> | Requirements:
> | =============
> | Either CMUCL (version 18b), SBCL (version 0.8.5) or 
> | Allegro CL (version 5.0 or later) is required.
> 
> No mention of SBCL there...
> 
> | * Allegro CL: Matlisp should compile on Linux, Solaris, Windows.
> |               In particular, Matlisp relies on the foreign
> |               function interface of Allegro CL.
> | 
> | * CMU CL: Matlisp should compile on Linux, Solaris.
> |           In particular, Matlisp relies on the foreign function
> |           interface of CMU CL and the built-in type 
> |           kernel::complex-double-float.
> | 
> | * SBCL: Matlisp should compile on Linux and MacOS X
> |         In particular, Matlisp relies on the foreign function
> |         interface of SBCL.
> \----
> 
> ...but that last paragraph is encouraging.  The detailed instructions
> include
> 
> /----
> | 4. Configure the system.
> |    o If you're using Allegro CL, specify --with-lisp=acl.
> |    o If you're using CMU CL, specify --with-lisp=cmucl
> |    o If you're using SBCL, specify --with-lisp=sbcl
> |    o Give the name of the Lisp executable via --with-lisp-exec=<name>
> |    o Run configure:
> | 
> |    ./configure --with-lisp=<lisp> --with-lisp-exec=<exec> --prefix=`pwd`
> \----
> 
> so I tried
> 
> # ./configure --with-lisp=sbcl --with-lisp-exec=/usr/bin/sbcl --prefix=`pwd`
> 
> only to be told
> 
> checking build system type... i686-pc-linux-gnu
> checking host system type... i686-pc-linux-gnu
> checking target system type... i686-pc-linux-gnu
> configure: error: Unknown Lisp system:  sbcl.  Supported systems are acl and cmucl
> 
>     Now, I know one plays with CVS versions at his own risk, and I'm
> probably being dense to boot, but there's enough there to suggest
> matlisp ought to work with SBCL that I thought it might be worth a
> note.  Will someone please enlighten me about the status of SBCL
> compatibility, or point out where I've gone wrong in the ./configure
> line?  Thanks!
> 
> Michael
> 

Re: [Matlisp-users] Configuration failure on Linux/SBCL

[Raymond T. <to...@rt...>]

>>>>> "Robbie" == Robbie Sedgewick <rd...@me...> writes:


    Robbie> Unfortunately, there were some problems with the patch (it got messed
    Robbie> up by the sourceforge mailing list) so it wasn't completely applied to
    Robbie> the cvs tree.  Parts of the patch were correctly applied, and others
    Robbie> not.

I would appreciate it if you could redo the patch and send it again.
I'll make sure it gets applied.

Ray

Re: [Matlisp-users] Configuration failure on Linux/SBCL

[Robert S. <rd...@me...>]

On May 21, 2004, at 8:51 AM, Raymond Toy wrote:

>>>>>> "Robbie" == Robbie Sedgewick <rd...@me...> writes:
>
>
> I would appreciate it if you could redo the patch and send it again.
> I'll make sure it gets applied.

Because of the earlier problem with the patch being too large for the 
sourceforge mailing list, I put the patch up on the web.

Its up at http://mercury.chem.pitt.edu/~rds/matlisp.sbcl.patch
the sources I generated the patch from are at
http://mercury.chem.pitt.edu/~rds/matlisp.sbcl.tar.gz
If there is any problem with the patch.

It is diffed off the current cvs so applying it shouldn't be a problem. 
  In addition to the sbcl support I also added a in-package statement to 
lib-src/cpoly/zeroin.lisp.  It's nice to have that in the file if you 
ask me.....

Let me know when you get the patch applied and I'll test it....

Thanks,
Robbie

Re: [Matlisp-users] Configuration failure on Linux/SBCL

[Raymond T. <to...@rt...>]

>>>>> "Robert" == Robert Sedgewick <rd...@me...> writes:

    Robert> Let me know when you get the patch applied and I'll test it....

Patch applied without problems.  I'm checking it in now, so I'd
appreciate it if you could test it out when you get a chance.

Ray

Re: [Matlisp-users] Configuration failure on Linux/SBCL

[Robbie S. <rd...@me...>]

Raymond Toy wrote:
> Patch applied without problems.  I'm checking it in now, so I'd
> appreciate it if you could test it out when you get a chance.
> 

Great, it seems to work here (sbcl on linux/x86), I'll try it on 
darwin/ppc when I get home, but it should work there too.

Now the only thing left to change is the website to let people know that 
it works with sbcl and on Mac OS X......


Thanks,
Robbie

Re: [Matlisp-users] Configuration failure on Linux/SBCL

[<dm...@ui...>]

Robbie Sedgewick <rd...@me...> writes:

> Hi,
>
> I did the port of matlisp to sbcl.

Thanks!

> Unfortunately, there were some problems with the patch (it got messed
> up by the sourceforge mailing list) so it wasn't completely applied to
> the cvs tree.  Parts of the patch were correctly applied, and others
> not.
>
> I have a copy of my sources up at
> http://mercury.chem.pitt.edu/~rds/matlisp.sbcl.tar.gz
>
> If you download these sources they should work out of the box. The
> only thing you might have to do is rebuild the configure script
> ("autoconf configure.in > configure").  After this the installation
> directions should work.

The command

    autoconf configure.in > configure

produced

    autoconf: Undefined macros:
    configure.in:259:AC_F77_FUNC(f77_name)

It's certainly possible this is a symptom of a broader configuration
problem, but I don't recall seeing any similar behavior in the past.

I then tried the existing configure script, but that led to

/----
| debugger invoked on a SB-FASL::INVALID-FASL-IMPLEMENTATION in thread 25305:
|   #<FILE-STREAM for "file \"/home/dmmcf/lisp/lib/matlisp.sbcl/packages.fasl\""
|     {98C8C69}> was compiled for implementation PPC, but this is a X86.
\----

which is hard to argue with. :-)

> Let me know if you have any problems.....

Oh, I can break anything.  Thanks for the help.

Michael

-- 
D. Michael McFarland
Department of Aerospace Engineering
University of Illinois at Urbana-Champaign
mailto:dm...@ui... https://netfiles.uiuc.edu/dmmcf/www/

Re: [Matlisp-users] Configuration failure on Linux/SBCL

[Robbie S. <rd...@me...>]

D. Michael McFarland wrote:

> 
> 
> The command
> 
>     autoconf configure.in > configure
> 
> produced
> 
>     autoconf: Undefined macros:
>     configure.in:259:AC_F77_FUNC(f77_name)
> 
> It's certainly possible this is a symptom of a broader configuration
> problem, but I don't recall seeing any similar behavior in the past.
> 

It probably has something to do with your version of autoconf.  I don't 
know much about it.


> I then tried the existing configure script, but that led to
> 
> /----
> | debugger invoked on a SB-FASL::INVALID-FASL-IMPLEMENTATION in thread 25305:
> |   #<FILE-STREAM for "file \"/home/dmmcf/lisp/lib/matlisp.sbcl/packages.fasl\""
> |     {98C8C69}> was compiled for implementation PPC, but this is a X86.
> \----
> 
> which is hard to argue with. :-)
> 

Well, if you got this far then the configure script is probably fine. 
This must have happened after you tried to make it, right?
Just remove all the .fasl files in the distribution and try it again. 
It should work.

--Robbie

> 
>>Let me know if you have any problems.....
> 
> 
> Oh, I can break anything.  Thanks for the help.
> 
> Michael
> 

Re: [Matlisp-users] Configuration failure on Linux/SBCL

[<dm...@ui...>]

Robbie Sedgewick <rd...@me...> writes:

> Well, if you got this far then the configure script is probably
> fine. This must have happened after you tried to make it, right?
> Just remove all the .fasl files in the distribution and try it
> again. It should work.

After a little more brute-force cleanup and a fresh attempt, things
seem to be working.  Thanks for the code (and your interest).

Best regards,
Michael

-- 
D. Michael McFarland
Department of Aerospace Engineering
University of Illinois at Urbana-Champaign
mailto:dm...@ui... https://netfiles.uiuc.edu/dmmcf/www/

Re: [Matlisp-users] Configuration failure on Linux/SBCL

[Raymond T. <to...@rt...>]

>>>>> "Robbie" == Robbie Sedgewick <rd...@me...> writes:

    Robbie> D. Michael McFarland wrote:

    >> The command
    >> autoconf configure.in > configure
    >> produced
    >> autoconf: Undefined macros:
    >> configure.in:259:AC_F77_FUNC(f77_name)
    >> It's certainly possible this is a symptom of a broader configuration
    >> problem, but I don't recall seeing any similar behavior in the past.
    >> 

    Robbie> It probably has something to do with your version of autoconf.  I
    Robbie> don't know much about it.

You need autoconf 2.50 or later, I think.  That macro was added around
that time.  Maybe a little earlier.

Ray

Re: [Matlisp-users] Configuration failure on Linux/SBCL

[<dm...@ui...>]

Raymond Toy <to...@rt...> writes:

> You need autoconf 2.50 or later, I think.  That macro was added around
> that time.  Maybe a little earlier.

It's all coming back to me now...well, what I ever knew of autoconf
is coming back to me....

It appears I have both versions 2.13 and 2.58 installed in /usr/bin,
while autoconf is a link to a wrapper that invokes one or the other:

dmmcf@ness% ls -l /usr/bin/autoconf*
    Apr 19 19:46 /usr/bin/autoconf -> ../lib/autoconf/ac-wrapper.pl
    Apr 19 19:46 /usr/bin/autoconf-2.13
    Apr 19 19:46 /usr/bin/autoconf-2.58

dmmcf@ness% autoconf --version
    Autoconf version 2.13

dmmcf@ness% WANT_AUTOCONF=2.5 autoconf --version
    autoconf (GNU Autoconf) 2.58

Sure enough,

    WANT_AUTOCONF=2.5 autoconf configure.in > configure

runs fine with Robbie's code.

Thanks, Ray.  I feel a little better now.

Cheers,
Michael