matlisp-users — General discussion list.

Re: [Matlisp-users] matlisp on Mac OS X (Intel)

[Jan R. <ja...@ry...>]

> Jan,
> I'm trying to get things working with Mac OS X on intel as well.  I  
> am also using the g77, gfortran, and gcc from the Mac HPC page on  
> sourceforge.  I've noticed that the default distribution from cvs  
> seems to have been built assuming that folks would be using Fink.   
> I'm definitely not and that's what's causing the link to fail for  
> me.  I've found that if you'll edit configure.in, search for "-L/sw/ 
> lib" down under the case for darwin you'll find some rules that  
> pertain to the problem at hand.  If you'll replace the line that says:
> 
> 	FLIBS="-L/sw/lib -lf2c -lSystem /usr/lib/bundle1.o"
> 
> with
> 
> 	FLIBS="-lSystem /usr/lib/bundle1.o"
> 
> and then save the file and run "autoconf configure.in > configure"  
> and then ./configure as usual, and then make, you should get through  
> the link.  This is about as far as I've gotten so far.  Which lisp  
> are you using?  I'm using SBCL.

This is a good bit of advice -- one thing I was missing was that I've
been trying to use the included configure script, instead of
regenerating it using autoconf. Running autoconf definitely improved
things.

libmatlisp.so does get built now -- but SBCL is unable to load it:

debugger invoked on a SIMPLE-ERROR:
  Error opening shared object "/Users/jwr/lisp/matlisp/lib/libmatlisp.so":
  dlopen(3) failed.

[10:40] tnuctip:/Users/jwr>file /Users/jwr/lisp/matlisp/lib/libmatlisp.so
/Users/jwr/lisp/matlisp/lib/libmatlisp.so: Mach-O bundle i386
[10:40] tnuctip:/Users/jwr>

I thought SBCL did support loading shared objects on Mac OS X?

AllegroCL seems to load the same library just fine.

--J.

> On Apr 20, 2007, at 4:50 AM, Jan Rychter wrote:
> 
> > Does anyone use matlisp on Mac OS X on an Intel CPU? If so, which
> > fortran compiler did you use? Which exact version and where did you  
> > get
> > it?
> >
> > I can't seem to get through the linking phase for BLAS/LAPACK. I tried
> > gfortran from the gcc 4.3 package from http://hpc.sourceforge.net/,  
> > and
> > now tried g77 from GNU Fortran (GCC) 3.4.0. They work fine until ld  
> > gets
> > invoked and then things break. One reason for the breakage is that
> > configure insists on trying to link in crt1.o into libraries, which
> > won't work, but even removing that doesn't get me anywhere.
> >
> > Since some people reported being able to get it to work with minor
> > adjustments, there must be something I'm doing wrong.
> >
> > Any hints?
> >
> > --J.

Re: [Matlisp-users] matlisp on Mac OS X (Intel)

[Enphilistor <enp...@ma...>]

Jan,

I'm trying to get things working with Mac OS X on intel as well.  I  
am also using the g77, gfortran, and gcc from the Mac HPC page on  
sourceforge.  I've noticed that the default distribution from cvs  
seems to have been built assuming that folks would be using Fink.   
I'm definitely not and that's what's causing the link to fail for  
me.  I've found that if you'll edit configure.in, search for "-L/sw/ 
lib" down under the case for darwin you'll find some rules that  
pertain to the problem at hand.  If you'll replace the line that says:

	FLIBS="-L/sw/lib -lf2c -lSystem /usr/lib/bundle1.o"

with

	FLIBS="-lSystem /usr/lib/bundle1.o"

and then save the file and run "autoconf configure.in > configure"  
and then ./configure as usual, and then make, you should get through  
the link.  This is about as far as I've gotten so far.  Which lisp  
are you using?  I'm using SBCL.

Good luck!

-- E


On Apr 20, 2007, at 4:50 AM, Jan Rychter wrote:

> Does anyone use matlisp on Mac OS X on an Intel CPU? If so, which
> fortran compiler did you use? Which exact version and where did you  
> get
> it?
>
> I can't seem to get through the linking phase for BLAS/LAPACK. I tried
> gfortran from the gcc 4.3 package from http://hpc.sourceforge.net/,  
> and
> now tried g77 from GNU Fortran (GCC) 3.4.0. They work fine until ld  
> gets
> invoked and then things break. One reason for the breakage is that
> configure insists on trying to link in crt1.o into libraries, which
> won't work, but even removing that doesn't get me anywhere.
>
> Since some people reported being able to get it to work with minor
> adjustments, there must be something I'm doing wrong.
>
> Any hints?
>
> --J.
>
>
> ---------------------------------------------------------------------- 
> ---
> This SF.net email is sponsored by DB2 Express
> Download DB2 Express C - the FREE version of DB2 express and take
> control of your XML. No limits. Just data. Click to get it now.
> http://sourceforge.net/powerbar/db2/
> _______________________________________________
> Matlisp-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matlisp-users

[Matlisp-users] matlisp on Mac OS X (Intel)

[Jan R. <ja...@ry...>]

Does anyone use matlisp on Mac OS X on an Intel CPU? If so, which
fortran compiler did you use? Which exact version and where did you get
it?

I can't seem to get through the linking phase for BLAS/LAPACK. I tried
gfortran from the gcc 4.3 package from http://hpc.sourceforge.net/, and
now tried g77 from GNU Fortran (GCC) 3.4.0. They work fine until ld gets
invoked and then things break. One reason for the breakage is that
configure insists on trying to link in crt1.o into libraries, which
won't work, but even removing that doesn't get me anywhere.

Since some people reported being able to get it to work with minor
adjustments, there must be something I'm doing wrong.

Any hints?

--J.

Re: [Matlisp-users] (use-package "MATLISP") doesn't work

[Raymond T. <ray...@er...>]

>>>>> "Tamas" == Tamas K Papp <Tamas> writes:

    Tamas> I start cmucl (latest version) and do (load "start.lisp") from the
    Tamas> directory where I put matlisp.

    Tamas> I try to load it with use-package, but get the following error message:

    Tamas> *  (use-package "MATLISP")


    Tamas> Error in function USE-PACKAGE:
    Tamas>    Use'ing package MATLISP results in name conflicts for these symbols:
    Tamas> (REAL HELP)
    Tamas>    [Condition of type SIMPLE-ERROR]

This is a known issue.  Matlisp defines a function REAL in the MATLISP
package.  This collides with the symbol REAL in COMMON-LISP.  If you
continue, you will lose the symbol REAL from the CL-USER package.

    Tamas> OTOH, in-package works fine:


    Tamas> * (in-package "MATLISP-USER")

This is the preferred way to access Matlisp.

    Tamas> #<The MATLISP-USER package, 1/9 internal, 0/2 external>

    Tamas> but only after I run (load "start.lisp").

The package won't exist until you load start.lisp.

    Tamas> Questions:
    Tamas> 1. do I need to load start.lisp every time I want to use matlisp?
    Tamas> 2. how can I make use-package work?

1. You can save a core file with Matlisp and everything loaded.  
2. Not sure, but you need to be careful because Matlisp defines it's
   own REAL function.

Ray

[Matlisp-users] (use-package "MATLISP") doesn't work

[Tamas K P. <tp...@Pr...>]

I start cmucl (latest version) and do (load "start.lisp") from the
directory where I put matlisp.

I try to load it with use-package, but get the following error message:

*  (use-package "MATLISP")


Error in function USE-PACKAGE:
   Use'ing package MATLISP results in name conflicts for these symbols:
(REAL HELP)
   [Condition of type SIMPLE-ERROR]

Restarts:
  0: [CONTINUE] Unintern the conflicting symbols in the COMMON-LISP-USER package.
  1: [ABORT   ] Return to Top-Level.

Debug  (type H for help)

(USE-PACKAGE "MATLISP"
             #<The COMMON-LISP-USER package, 8/9 internal, 0/9 external>)
Source: Error finding source:
Error in function DEBUG::GET-FILE-TOP-LEVEL-FORM:  Source file no longer exists:
  target:code/package.lisp.
0]


OTOH, in-package works fine:


* (in-package "MATLISP-USER")

#<The MATLISP-USER package, 1/9 internal, 0/2 external>

but only after I run (load "start.lisp").

Questions:
1. do I need to load start.lisp every time I want to use matlisp?
2. how can I make use-package work?

Sorry if the question is dumb, I am new to Lisp.

Thanks,

Tamas

Re: [Matlisp-users] could not install matlisp

[Raymond T. <ray...@er...>]

>>>>> "Tamas" == Tamas Papp <tk...@gm...> writes:

    Tamas> 1.  (*) text/plain          ( ) text/html           
    Tamas> Raymond,

    Tamas> Thanks, the CVS version worked.  I am using CMUCL, version 19d-20061116-1.

Ok, glad that worked.

    Tamas> Is there some kind of overview to get me started?  I found the examples in
    Tamas> the README, but nothing else.  I can of course look at the help of all
    Tamas> commands, but I wonder if there is some other help document/manual, a but
    Tamas> more structured.

Unfortunately not.  About the only documentation there is is at
matlisp.sourceforge.net, and the (help) command.  Most of the matrix
commands use exactly the same name as LAPACK/BLAS, so that makes it
easier to figure out what they do.  

I guess the only thing not really documented is that you can do
something like:

(defvar *m* [1 2 3; 4 5 6])

to create a matrix 

    1 2 3
    4 5 6

Hope this helps a little,

Ray

Re: [Matlisp-users] could not install matlisp

[Tamas P. <tk...@gm...>]

Raymond,

Thanks, the CVS version worked.  I am using CMUCL, version 19d-20061116-1.

Is there some kind of overview to get me started?  I found the examples in
the README, but nothing else.  I can of course look at the help of all
commands, but I wonder if there is some other help document/manual, a but
more structured.

Thanks,

Tamas


On 4/2/07, Raymond Toy <ray...@er...> wrote:
>
> >>>>> "Tamas" == Tamas Papp <tk...@gm...> writes:
>
>     Tamas> 1.  (*) text/plain          ( ) text/html
>     Tamas> Hi,
>
>     Tamas> I tried to compile matlisp, but make failed (see error messages
> below).  I
>     Tamas> also
>     Tamas> included the output from configure.  Sorry if it is something
> obvious,
>     Tamas> I am a Lisp newbie.
>
> What Lisp are you using?  In addition, I suggest you use the CVS
> version.  The version you have is a bit old (although there probably
> isn't much differen).
>
> Ray
>
>
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share
> your
> opinions on IT & business topics through brief surveys-and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> Matlisp-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matlisp-users
>

Re: [Matlisp-users] could not install matlisp

[Raymond T. <ray...@er...>]

>>>>> "Tamas" == Tamas Papp <tk...@gm...> writes:

    Tamas> 1.  (*) text/plain          ( ) text/html           
    Tamas> Hi,

    Tamas> I tried to compile matlisp, but make failed (see error messages below).  I
    Tamas> also
    Tamas> included the output from configure.  Sorry if it is something obvious,
    Tamas> I am a Lisp newbie.

What Lisp are you using?  In addition, I suggest you use the CVS
version.  The version you have is a bit old (although there probably
isn't much differen).

Ray

[Matlisp-users] could not install matlisp

[Tamas P. <tk...@gm...>]

Hi,

I tried to compile matlisp, but make failed (see error messages below).  I
also
included the output from configure.  Sorry if it is something obvious,
I am a Lisp newbie.

Thanks,

Tamas


% ./configure
checking build system type... i686-pc-linux-gnu
checking host system type... i686-pc-linux-gnu
checking target system type... i686-pc-linux-gnu
checking for ranlib... ranlib
checking for gcc... gcc
checking for C compiler default output... a.out
checking whether the C compiler works... yes
checking whether we are cross compiling... no
checking for executable suffix...
checking for object suffix... o
checking whether we are using the GNU C compiler... yes
checking whether gcc accepts -g... yes
checking whether gcc and cc understand -c and -o together... yes
checking for g77... g77
checking whether we are using the GNU Fortran 77 compiler... yes
checking whether g77 accepts -g... yes
checking whether g77 understand -c and -o together... yes
checking whether we are using the GNU Fortran 77 compiler... (cached) yes
checking whether g77 accepts -g... (cached) yes
checking how to get verbose linking output from g77... -v
checking for Fortran 77 libraries...  -L/usr/lib/gcc/i486-linux-gnu/3.4.6
-L/usr/lib/gcc/i486-linux-gnu/3.4.6/../../../../lib
-L/usr/lib/gcc/i486-linux-gnu/3.4.6/../../.. -L/lib/../lib
-L/usr/lib/../lib+-lfrtbegin -lg2c -lm -lgcc_s
checking for dummy main to link with Fortran 77 libraries... none
checking for Fortran 77 name-mangling scheme... lower case, underscore,
extra+underscore
configure: creating ./config.status
config.status: creating matlisp.mk
config.status: creating lib/lazy-loader.lisp
config.status: creating src/f77-mangling.lisp
% make
make -f matlisp.mk
make[1]: Entering directory
`/home/tpapp/doc/lisp/matlisp-2_0beta-2003-10-14''(progn (load "start.lisp
"))'
make[1]: (progn (load "start.lisp")): Command not found
make[1]: *** [all] Error 127
make[1]: Leaving directory `/home/tpapp/doc/lisp/matlisp-2_0beta-2003-10-14'
make: *** [all] Error 2

Re: [Matlisp-users] third party fortran libraries and REAL-MATRIX type

[Raymond T. <ray...@er...>]

>>>>> "Dmitry" == Dmitry V Gorbatovsky <gor...@gm...> writes:

    Dmitry> Hello,
    Dmitry> I am try to figure out:
    Dmitry> 1. how to use Matlisp-Fortran interface to load third party Fortran libraries 

You need to tell matlisp how to call the Fortran function.  You can
see examples in, for example, src/blas.lisp.

Ask if you need some help with that.

    Dmitry> and 
    Dmitry> 2.how to pass #<MATLISP:REAL-MATRIX type to third party Fortran routines.

Once you've done the above, you're almost there.  Look in
src/axpy.lisp for an example.

If you need more help, tell us one of the functions you want to call,
and we'll help you define the necessary things to call it.

Ray

[Matlisp-users] third party fortran libraries and REAL-MATRIX type

[Dmitry V. G. <gor...@gm...>]

Hello,
I am try to figure out:
1. how to use Matlisp-Fortran interface to load third party Fortran libraries 
and 
2.how to pass #<MATLISP:REAL-MATRIX type to third party Fortran routines.

Few line guide would be sufficient.
I would appreciate any help .

Thanks, Dmitry

Re: [Matlisp-users] fft -> SIGSEGV

[Raymond T. <ray...@er...>]

>>>>> "Thibault" == Thibault Langlois <tl...@di...> writes:

    Thibault> Hello,
    Thibault> I don't know if something is wrong with my matlisp installation but I
    Thibault> get a segmentation violation with this little test involving fft:

    Thibault> I am using cmucl-19c, cvs matlisp.

[snip]

Can you do other matrix operations besides FFT?  Do they work?

Your test example works fine for me with CMUCL CVS for sparc.
Unfortunately, I don't have any Linux box with a Fortran compiler to
test this out.

Ray

[Matlisp-users] fft -> SIGSEGV

[Thibault L. <tl...@di...>]

Hello,

I don't know if something is wrong with my matlisp installation but I
get a segmentation violation with this little test involving fft:

I am using cmucl-19c, cvs matlisp.

(in-package :matlisp)

(print (lisp-implementation-type))
(print (lisp-implementation-version))
(print *features*)

(defconstant 2pi (* 2 pi))

(defparameter *data-1*
  (make-real-matrix
   (loop with a = 1
         with f = 0.25
         with sr = 1
         with phi = 0
         for n from 0 below 64
         for x = (* a (cos (+ (* 2pi n f sr) phi)))
         collect x)))

(fft *data-1*)

wok$ ~/comp/lang/lisp/cmucl-19c/bin/lisp
-core /usr/local/src/matlisp/matlisp.core -load
"/home/tl/Projects/sandbox/fft.lisp"
; Loading #P"/home/tl/init.lisp".
Warning:  Declaring *LOAD-OBJECT-TYPES* special.
;; Loading #P"/home/tl/comp/lang/lisp/code/asdf/asdf.x86f".
;; Loading #P"/home/tl/root/lisp/pdoc.lisp".
; Loading #P"/home/tl/Projects/sandbox/fft.lisp".

"CMU Common Lisp"
"19c (19C)"
(:ASDF :TL :MK-DEFSYSTEM :MATLISP :GERDS-PCL :PCL-STRUCTURES
:PORTABLE-COMMONLOOPS :PCL :CMU19 :CMU19C :PYTHON :CONSERVATIVE-FLOAT-TYPE
:MODULAR-ARITH :MP :X86 :LINKAGE-TABLE :RELATIVE-PACKAGE-NAMES :LINUX :GLIBC2
:UNIX :RANDOM-MT19937 :GENCGC :PENTIUM :I486 :HASH-NEW :HEAP-OVERFLOW-CHECK


Error in function UNIX::SIGSEGV-HANDLER:  Segmentation Violation at
#xB7F43775.
   [Condition of type SIMPLE-ERROR]

Restarts:
  0: [CONTINUE] Return NIL from load of
"/home/tl/Projects/sandbox/fft.lisp".
  1: [ABORT   ] Skip remaining initializations.

Debug  (type H for help)

(UNIX::SIGSEGV-HANDLER #<unused-arg>
                       #<unused-arg>
                       #.(SYSTEM:INT-SAP #x3FFFC5B8))
Source: Error finding source:
Error in function DEBUG::GET-FILE-TOP-LEVEL-FORM:  Source file no longer
exists:  target:code/signal.lisp.
0]


-- 
Thibault Langlois <tl...@di...>
FCUL / DI

Re: [Matlisp-users] Using --with-atlas with sbcl

[Raymond T. <ray...@er...>]

>>>>> "jao" == Jose A Ortega Ruiz <ja...@gn...> writes:

    jao> Hi. As a result of a conversation with rtoy at #lisp, I've noticed
    jao> that lazy-loader.lisp is not loading libblas&c. when configure
    jao> --with-atlas is used with sbcl. This quick hack of
    jao> load-blas-&-lapack-libraries works for me:

Thanks for the patch.

I think that to make this work right, I'll have to add another option
--enable-shared-atlas because not all platforms have shared atlas
libraries.   

I'll have to play around a bit with this, though.  I've kind of
forgotten exactly what libs are needed to build matlisp with ATLAS.

Ray

Re: [Matlisp-users] Minor change to compile Matlisp on Darwin with SBCL

[Raymond T. <ray...@er...>]

>>>>> "Matthew" == Matthew West <we...@st...> writes:

    Matthew> To compile on OS X with SBCL I needed the patch below, fixing what  
    Matthew> appears to be a simple missing line.

Sorry for the delay.  I've applied this patch (finally).  

Thanks!

Ray

[Matlisp-users] Using --with-atlas with sbcl

[Jose A. O. R. <ja...@gn...>]

Hi. As a result of a conversation with rtoy at #lisp, I've noticed
that lazy-loader.lisp is not loading libblas&c. when configure
=2D-with-atlas is used with sbcl. This quick hack of
load-blas-&-lapack-libraries works for me:

#+:sbcl
(defun load-blas-&-lapack-libraries ()
  #+@ATLAS_P@ (sb-alien:load-shared-object "@ATLAS_DIR@/libblas.so")
  #+@ATLAS_P@ (sb-alien:load-shared-object "@ATLAS_DIR@/liblapack.so")
  #(not @ATLAS_P@) (sb-alien:load-shared-object "matlisp:lib;libmatlisp.so"=
))

HTH,
jao
=2D-=20
If people do not believe that mathematics is simple, it is only
because they do not realize how complicated life is.
=2D John von Neumann

[Matlisp-users] Minor change to compile Matlisp on Darwin with SBCL

[Matthew W. <we...@st...>]

Hi all,

Let me start by saying thanks for such a nice package! I've been  
using it for some time as the basis for some simple optimization code  
and it's performing very well.

To compile on OS X with SBCL I needed the patch below, fixing what  
appears to be a simple missing line.

For anyone who's interested, Matlisp works out-of-the-box (subject to  
the single-line change below) with OS X 10.4, SBCL 0.9.15, CVS  
Matlisp as of 2006-08-12, and g77 3.4.3 (from fink).

Matt.


--- configure.in.old    2006-08-12 22:05:48.000000000 -0700
+++ configure.in        2006-08-12 22:06:08.000000000 -0700
@@ -357,6 +357,7 @@
         ;;
    *-*-darwin*)
          LD="ld"
+       MATLIB="libmatlisp.so"

         # Cyrus Harmon says -lcc_dynamic isn't needed for Mac OS X
         # 10.4.  But it's needed (apparently) for earlier versions.

Re: [Matlisp-users] (MAKE-COMPLEX-MATRIX ...) SIGSEGV Problem

[Raymond T. <ray...@er...>]

>>>>> "Michael" == Michael A Koerber <ma...@ll...> writes:

    Michael> I forgot I even reported this :-)
    Michael> AFAIK CMUCL works fine for other things.  But I haven't "stressed" it
    Michael> too much.  Once I got the SIGV I was dead in my tracks since this
    Michael> "locks out" several years of code that I have developed.

    Michael> I'm pushing to get permission to use FreeBSD since the OSs offered
    Michael> to me for my use are "feature-starved".  If that fails, I may need
    Michael> to find a solution to the SEGV problem.

Let me know how it goes.  I'll also try to find some linux box that
has a fortran compiler.

You may want to try 19c too, or a recent snapshot.  I think there were
some fixes there for some Linux issues.  Something about memory map
randomization or something.

Sorry about the troubles.

Ray

Re: [Matlisp-users] compiling MatLisp on FC5?

[Paul B. <pb...@dw...>]

Michael,

(MAKE-COMPLEX-MATRIX 10) works fine for me -- but then, again, I
switched to sbcl, as described here:
http://groups.google.ca/group/comp.lang.lisp/browse_frm/thread/e92690b0d496bdd0

Paul B.

On Thu, 2006-06-08 at 10:13 -0400, Michael A. Koerber wrote:
> Paul,
> 
> If/when you get it to compile, execute (make-complex-matrix N) with N
> some number greater than 2.  I haven't had time to investigate the
> problem (nor report it), but under SLES-9 and RHE-3 (?) I get SEGV
> problems.  I might suspect the same for for FC-x.
> 
> mike

Re: [Matlisp-users] (MAKE-COMPLEX-MATRIX ...) SIGSEGV Problem

[Michael A. K. <ma...@ll...>]

I forgot I even reported this :-)

AFAIK CMUCL works fine for other things.  But I haven't "stressed" it
too much.  Once I got the SIGV I was dead in my tracks since this
"locks out" several years of code that I have developed.

I'm pushing to get permission to use FreeBSD since the OSs offered
to me for my use are "feature-starved".  If that fails, I may need
to find a solution to the SEGV problem.

mike


At Wed, 07 Jun 2006 09:18:41 -0400,
Raymond Toy <ray...@er...> wrote:
> 
> >>>>> "Michael" == Michael A Koerber <ma...@ll...> writes:
> 
>     Michael> All,
>     Michael>   I'm using "matlisp-2.0beta-2003-10-14.tar.gz" with CMUCL 19B.  Under
>     Michael> FreeBSD 5.4 and 6.0 all is well.
> 
>     Michael>   Under SUSE (SLES-9) and Red Hat Enterprise Linux WS, Version 3 I get
>     Michael> segment violation (see below).  (A long time ago I believe I saw
>     Michael> something like this that was attributed to shared object library
>     Michael> offsets changing.)  
> 
>     Michael>   Has this been noted before...is there a fix?
> 
> Sorry for the long delay.  I have Suse 9.0 but don't have a compatible
> f77 compiler to test matlisp with.  (Suse 9.3 and later won't install
> on my machine either because it can't boot from any of the install
> CDs.  Fooey.)
> 
> I don't have any problem with matlisp (CVS) with Solaris.
> 
> Does cmucl work ok for other things besides matlisp?
> 
> Ray
> 
> 
> 
> _______________________________________________
> Matlisp-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matlisp-users
> 

Re: [Matlisp-users] compiling MatLisp on FC5?

[Michael A. K. <ma...@ll...>]

Paul,

If/when you get it to compile, execute (make-complex-matrix N) with N
some number greater than 2.  I haven't had time to investigate the
problem (nor report it), but under SLES-9 and RHE-3 (?) I get SEGV
problems.  I might suspect the same for for FC-x.

mike

At Tue, 06 Jun 2006 16:51:49 -0700,
Paul Bunyk <pb...@dw...> wrote:
> 
> Hello All!
> 
> I'm wondering if anyone managed to successfully compile (and run!)
> MatLisp on FC5 (that's Fedora Core 5 for non-Linux folks :-) )? 
> 
> I'm using stock CMU Common Lisp 19c Fedora Extras release 4.fc5.
> 
> MatLisp configured/compiled just fine (all I had to change was adding
> (load "packages") in the beginning of start.lisp -- otherwise it
> complained about unknown "F2CL-LIB" package), but when I try any
> non-trivial matrix operation (such as m+, which, I think is the simplest
> one making an actual Fortran call) it gets into an infinite loop
> (breakable with ^C though). When I run it under gdb it seems to be
> getting SIGSEGVs when (unsuccessfully) trying to load a symbol in
> dl_-something. 
> 
> It worked fine for me on FC4. One of the obvious differences is that FC4
> used gcc 4.0, while FC5 uses 4.1. I've tried downgrading C compiler to
> gcc32, upgrading Fortran compiler to gfortran (failed in even more
> spectacular way!), nothing helps.
> 
> CMUCL docs mention that "There are problems running CMUCL on certain
> kernels that are built with address-space randomization techniques, such
> as certain Fedora kernels. If you run into problems, please use a
> vanilla kernel." -- I have 2.6.16-1.2111_FC5 and I *assume* that Fedora
> folks would make their CMUCL rpm work with their stock kernel -- is it
> too much to assume? :-)
> 
> Thanks in advance!
> 
> Paul Bunyk
> 
> 
> -- 
> Paul Bunyk <pb...@dw...>
> D-Wave: The Quantum Computing Company
> 
> 
> 
> _______________________________________________
> Matlisp-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matlisp-users
> 

Re: [Matlisp-users] compiling MatLisp on FC5?

[Raymond T. <ray...@er...>]

>>>>> "Paul" == Paul Bunyk <pb...@dw...> writes:

    Paul> Hello All!
[snip]
    Paul> CMUCL docs mention that "There are problems running CMUCL on certain
    Paul> kernels that are built with address-space randomization techniques, such
    Paul> as certain Fedora kernels. If you run into problems, please use a
    Paul> vanilla kernel." -- I have 2.6.16-1.2111_FC5 and I *assume* that Fedora
    Paul> folks would make their CMUCL rpm work with their stock kernel -- is it
    Paul> too much to assume? :-)

I can't help much because I don't have FC5 and I can't even build
matlisp on my linux box.

However, you may want to ask on the cmucl-help mailing list to see if
anyone has cmucl working on FC5.  I suspect it does, but I don't
really know.

Ray

Re: [Matlisp-users] (MAKE-COMPLEX-MATRIX ...) SIGSEGV Problem

[Raymond T. <ray...@er...>]

>>>>> "Michael" == Michael A Koerber <ma...@ll...> writes:

    Michael> All,
    Michael>   I'm using "matlisp-2.0beta-2003-10-14.tar.gz" with CMUCL 19B.  Under
    Michael> FreeBSD 5.4 and 6.0 all is well.

    Michael>   Under SUSE (SLES-9) and Red Hat Enterprise Linux WS, Version 3 I get
    Michael> segment violation (see below).  (A long time ago I believe I saw
    Michael> something like this that was attributed to shared object library
    Michael> offsets changing.)  

    Michael>   Has this been noted before...is there a fix?

Sorry for the long delay.  I have Suse 9.0 but don't have a compatible
f77 compiler to test matlisp with.  (Suse 9.3 and later won't install
on my machine either because it can't boot from any of the install
CDs.  Fooey.)

I don't have any problem with matlisp (CVS) with Solaris.

Does cmucl work ok for other things besides matlisp?

Ray

[Matlisp-users] compiling MatLisp on FC5?

[Paul B. <pb...@dw...>]

Hello All!

I'm wondering if anyone managed to successfully compile (and run!)
MatLisp on FC5 (that's Fedora Core 5 for non-Linux folks :-) )? 

I'm using stock CMU Common Lisp 19c Fedora Extras release 4.fc5.

MatLisp configured/compiled just fine (all I had to change was adding
(load "packages") in the beginning of start.lisp -- otherwise it
complained about unknown "F2CL-LIB" package), but when I try any
non-trivial matrix operation (such as m+, which, I think is the simplest
one making an actual Fortran call) it gets into an infinite loop
(breakable with ^C though). When I run it under gdb it seems to be
getting SIGSEGVs when (unsuccessfully) trying to load a symbol in
dl_-something. 

It worked fine for me on FC4. One of the obvious differences is that FC4
used gcc 4.0, while FC5 uses 4.1. I've tried downgrading C compiler to
gcc32, upgrading Fortran compiler to gfortran (failed in even more
spectacular way!), nothing helps.

CMUCL docs mention that "There are problems running CMUCL on certain
kernels that are built with address-space randomization techniques, such
as certain Fedora kernels. If you run into problems, please use a
vanilla kernel." -- I have 2.6.16-1.2111_FC5 and I *assume* that Fedora
folks would make their CMUCL rpm work with their stock kernel -- is it
too much to assume? :-)

Thanks in advance!

Paul Bunyk


-- 
Paul Bunyk <pb...@dw...>
D-Wave: The Quantum Computing Company

[Matlisp-users] (MAKE-COMPLEX-MATRIX ...) SIGSEGV Problem

[Michael A. K. <ma...@ll...>]

All,

  I'm using "matlisp-2.0beta-2003-10-14.tar.gz" with CMUCL 19B.  Under
FreeBSD 5.4 and 6.0 all is well.

  Under SUSE (SLES-9) and Red Hat Enterprise Linux WS, Version 3 I get
segment violation (see below).  (A long time ago I believe I saw
something like this that was attributed to shared object library
offsets changing.)  

  Has this been noted before...is there a fix?

tnx

mike


;; CMU Common Lisp 19b (19B), running on meatball.llan.ll.mit.edu
;; With core: /home/mak/bin/matlisp.core
;; Dumped on: Tue, 2006-05-09 10:13:30-04:00 on meatball.llan.ll.mit.edu
;; See <http://www.cons.org/cmucl/> for support information.
;; Loaded subsystems:
;;     Python 1.1, target Intel x86
;;     CLOS based on Gerd's PCL 2004/04/14 03:32:47
;;     MATLISP/Pre 2.0
;; 
;; * (m:make-complex-matrix 3)
;; 
;; 
;; 
;; Error in function UNIX::SIGSEGV-HANDLER:  Segmentation Violation at #xB75F4627.
;;    [Condition of type SIMPLE-ERROR]
;; 
;; Restarts:
;;   0: [ABORT] Return to Top-Level.
;; 
;; Debug  (type H for help)
;; 
;; (UNIX::SIGSEGV-HANDLER #<unused-arg>
;;                        #<unused-arg>
;;                        #.(SYSTEM:INT-SAP #x3FFFC834))
;; Source: Error finding source: 
;; Error in function DEBUG::GET-FILE-TOP-LEVEL-FORM:  Source file no longer exists:
;;   target:code/signal.lisp.
;; 0]