Regards, I've uploaded version 1.0b of Matlisp,
available from http://matlisp.sourceforge.net.
Version 1.0b supports CMUCL on Linux/Solaris (18b or later)
and Allegro CL on Linux/Solaris/MS Windows (5.0 or later).
This version also interfaces to DFFTPACK (except on windows).
Best,
Tunc
Tunc Simsek <······@robotics.eecs.berkeley.edu> writes:
> Regards, I've uploaded version 1.0b of Matlisp,
> available from http://matlisp.sourceforge.net.
first i want to say, thanks! common-lisp has good general math
functions but until now has lacked a good matrix package.
> Version 1.0b supports CMUCL on Linux/Solaris (18b or later)
hmm. i've been trying to get matlisp up and running on hp-ux with
cmucl 18a. it craps out in the loading of start.lisp. i get some
errors about logical paths. i am not at work, so i can't paste the
exact message. however, i will do so tomorrow.
what about using the ATLAS tuned BLAS? it has sped up octave for me.
is there any interest in this for matlisp?
> and Allegro CL on Linux/Solaris/MS Windows (5.0 or later).
> This version also interfaces to DFFTPACK (except on windows).
>
> Best,
> Tunc
--
J o h a n K u l l s t a m
[········@ne.mediaone.net]
Don't Fear the Penguin!
Hmm. These logical pathnames have been on my case for a while. I
thought
I had them figured out, guess not.
Johan Kullstam wrote:
>
> Tunc Simsek <······@robotics.eecs.berkeley.edu> writes:
>
> > Regards, I've uploaded version 1.0b of Matlisp,
> > available from http://matlisp.sourceforge.net.
>
> first i want to say, thanks! common-lisp has good general math
> functions but until now has lacked a good matrix package.
>
> > Version 1.0b supports CMUCL on Linux/Solaris (18b or later)
>
> hmm. i've been trying to get matlisp up and running on hp-ux with
> cmucl 18a. it craps out in the loading of start.lisp. i get some
> errors about logical paths. i am not at work, so i can't paste the
> exact message. however, i will do so tomorrow.
I think, as Ray said, that there are serious differences between
CMU 18a and CMU18b. First, there seems to be differences in the
logical pathnames. Second, it has been suggested that earlier
versions of CMU does not have the type kernel::complex-double-float.
Can you check if 18a has this, if not, that will also be an obstacle
in your installation.
Please report any bugs (even for older versions of CMU) to
matlisp.sourceforge.net.
>
> what about using the ATLAS tuned BLAS? it has sped up octave for me.
> is there any interest in this for matlisp?
Well, I've found ATLAS very impressive. There is a commented section
in lib/lazy-loader.lisp which shows my personal configuration of
ATLAS. The general makefile will build for the distributed fortran
sources. You may build as suggested, then edit lib/lazy-loader.lisp
manually to load ATLAS instead of BLAS (and if you're on Solaris, you
should also check if you have sunperf libraries).
>
> > and Allegro CL on Linux/Solaris/MS Windows (5.0 or later).
> > This version also interfaces to DFFTPACK (except on windows).
> >
> > Best,
> > Tunc
>
> --
> J o h a n K u l l s t a m
> [········@ne.mediaone.net]
> Don't Fear the Penguin!
>>>>> "Johan" == Johan Kullstam <········@ne.mediaone.net> writes:
Johan> hmm. i've been trying to get matlisp up and running on hp-ux with
Johan> cmucl 18a. it craps out in the loading of start.lisp. i get some
Johan> errors about logical paths. i am not at work, so i can't paste the
Logical pathname support in CMUCL got much better in later versions of
CMUCL. I don't remember if 18a is a "later version".
Johan> exact message. however, i will do so tomorrow.
Cool.
Johan> what about using the ATLAS tuned BLAS? it has sped up octave for me.
Johan> is there any interest in this for matlisp?
I've done this and so has Tunc. It was very much faster. You
basically change where standard BLAS gets loaded with the ATLAS
version.
Ray
Raymond Toy <···@rtp.ericsson.se> writes:
> >>>>> "Johan" == Johan Kullstam <········@ne.mediaone.net> writes:
>
>
> Johan> hmm. i've been trying to get matlisp up and running on hp-ux with
> Johan> cmucl 18a. it craps out in the loading of start.lisp. i get some
> Johan> errors about logical paths. i am not at work, so i can't paste the
>
> Logical pathname support in CMUCL got much better in later versions of
> CMUCL. I don't remember if 18a is a "later version".
>
> Johan> exact message. however, i will do so tomorrow.
>
> Cool.
ok you asked for it ;-)
from the start you can see it's 18a.
CMU Common Lisp 18a, running on edlhp221
Send bug reports and questions to your local CMU CL maintainer,
or to ·······@mikemac.com,
or to ··········@cons.org. (prefered)
Loaded subsystems:
Python 1.0, target HP-PA 1.1
CLOS based on PCL version: September 16 92 PCL (f)
i load up defsystem
* (load "defsystem")
; Loading #p"/var/tmp/matlisp/defsystem.hpf".
Warning: MAKE also exports the following symbols:
(MAKE:UNDEFSYSTEM MAKE:SYSTEM-SOURCE-SIZE MAKE:*MULTIPLE-LISP-SUPPORT*
MAKE:*CENTRAL-REGISTRY* MAKE:ALLEGRO-MAKE-SYSTEM-FASL
MAKE:*DEFSYSTEM-VERSION* MAKE:AFS-SOURCE-DIRECTORY
MAKE:*SOURCE-PATHNAME-DEFAULT* MAKE:FILES-IN-SYSTEM MAKE:DEFSYSTEM
MAKE:EDIT-SYSTEM MAKE:DEFINED-SYSTEMS MAKE:DESCRIBE-SYSTEM
MAKE:*COMPILE-DURING-LOAD* MAKE:FILES-WHICH-NEED-COMPILATION
MAKE:SOFTWARE-TYPE-TRANSLATION MAKE:*FILES-MISSING-IS-AN-ERROR*
MAKE:*DONT-REDEFINE-REQUIRE* MAKE:*BINARY-PATHNAME-DEFAULT*
MAKE:*BIN-SUBDIR* MAKE:OPERATE-ON-SYSTEM MAKE:FIND-SYSTEM
MAKE:*RELOAD-SYSTEMS-FROM-DISK* MAKE:*MINIMAL-LOAD* MAKE:OOS
MAKE:AFS-BINARY-DIRECTORY MAKE:DEFINE-LANGUAGE MAKE:HARDCOPY-SYSTEM
MAKE:LOAD-SYSTEM MAKE:COMPILER-TYPE-TRANSLATION MAKE:CLEAN-SYSTEM
MAKE:COMPILE-SYSTEM MAKE:MACHINE-TYPE-TRANSLATION MAKE:MAKE-SYSTEM-TAG-TABLE)
T
* (load "start.lisp")
; Loading #p"/var/tmp/matlisp/start.lisp".
[GC threshold exceeded with 2,850,824 bytes in use. Commencing GC.]
[GC completed with 761,400 bytes retained and 2,089,424 bytes freed.]
[GC will next occur when at least 2,761,400 bytes are in use.]
;; Loading #p"/var/tmp/matlisp/system.dcl".
Warning: This function is undefined:
DEFLOGICALPATH
Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER:
#() is not of type (OR CONS
BASE-STRING
(MEMBER NIL :UNSPECIFIC :WILD)
COMMON-LISP::PATTERN)
Restarts:
0: [CONTINUE] Return NIL from load of "system.dcl".
1: Return NIL from load of "start.lisp".
2: [ABORT ] Return to Top-Level.
Debug (type H for help)
("DEFUN MAKE-PATHNAME" 469762728 8)[:OPTIONAL]
Source: Error finding source:
Error in function DEBUG::GET-FILE-TOP-LEVEL-FORM: Source file no longer exists:
target:code/pathname.lisp.
0] q
i am finding cmucl to be hard to work with regarding finding where
errors occur in a large source file. do you have any hints as to how
to jump to error using ilisp?
i am not sure about the kernel::complex-double-float.
how do i tell if this exists?
thanks in advance.
--
J o h a n K u l l s t a m
[········@ne.mediaone.net]
edlhp
>>>>> "Johan" == Johan Kullstam <········@ne.mediaone.net> writes:
Johan> ok you asked for it ;-)
Johan> from the start you can see it's 18a.
[snip]
Johan> Restarts:
Johan> 0: [CONTINUE] Return NIL from load of "system.dcl".
Johan> 1: Return NIL from load of "start.lisp".
Johan> 2: [ABORT ] Return to Top-Level.
Johan> Debug (type H for help)
Johan> ("DEFUN MAKE-PATHNAME" 469762728 8)[:OPTIONAL]
Johan> Source: Error finding source:
Johan> Error in function DEBUG::GET-FILE-TOP-LEVEL-FORM: Source file no longer exists:
Johan> target:code/pathname.lisp.
Johan> 0] q
Johan> i am finding cmucl to be hard to work with regarding finding where
Johan> errors occur in a large source file. do you have any hints as to how
Johan> to jump to error using ilisp?
I don't know how to do that with ilisp. However, at the debug prompt,
you could type "back" to get a backtrace that might point you to the
offending code. You can go down the stack (stacks grow up on CMUCL)
and examine local variables to help. For more info, type H as
suggested.
Johan> i am not sure about the kernel::complex-double-float.
Johan> how do i tell if this exists?
If your version supports unboxed (complex double-float), I think you
have kernel::complex-double-float. There used to be a feature
:complex-float or something like that to indicate this, but later
versions don't have it because all ports support it.
Alternatively, try
(defmethod f ((f kernel::complex-double-float))
)
and see if CMUCL likes it.
Ray
ok, I got 18a for linux and tried but could not get matlisp to load.
the foreign function interface craps out (some segv error)
For your other questions:
the kernel::complex-double-float is not supported in 18a (apparently)
don't load your own defsystem.lisp (which may be an old version
causing the problems in the pathnames).
For now I would suggest getting a later version of cmu to work with.
Tunc
Johan Kullstam wrote:
>
> Raymond Toy <···@rtp.ericsson.se> writes:
>
> > >>>>> "Johan" == Johan Kullstam <········@ne.mediaone.net> writes:
> >
> >
> > Johan> hmm. i've been trying to get matlisp up and running on hp-ux with
> > Johan> cmucl 18a. it craps out in the loading of start.lisp. i get some
> > Johan> errors about logical paths. i am not at work, so i can't paste the
> >
> > Logical pathname support in CMUCL got much better in later versions of
> > CMUCL. I don't remember if 18a is a "later version".
> >
> > Johan> exact message. however, i will do so tomorrow.
> >
> > Cool.
>
> ok you asked for it ;-)
>
> from the start you can see it's 18a.
>
> CMU Common Lisp 18a, running on edlhp221
> Send bug reports and questions to your local CMU CL maintainer,
> or to ·······@mikemac.com,
> or to ··········@cons.org. (prefered)
>
> Loaded subsystems:
> Python 1.0, target HP-PA 1.1
> CLOS based on PCL version: September 16 92 PCL (f)
>
> i load up defsystem
>
don't do that if you're going to use start.lisp (this loads the provided
defsystem
by default)
> * (load "defsystem")
> ; Loading #p"/var/tmp/matlisp/defsystem.hpf".
> Warning: MAKE also exports the following symbols:
> (MAKE:UNDEFSYSTEM MAKE:SYSTEM-SOURCE-SIZE MAKE:*MULTIPLE-LISP-SUPPORT*
> MAKE:*CENTRAL-REGISTRY* MAKE:ALLEGRO-MAKE-SYSTEM-FASL
> MAKE:*DEFSYSTEM-VERSION* MAKE:AFS-SOURCE-DIRECTORY
> MAKE:*SOURCE-PATHNAME-DEFAULT* MAKE:FILES-IN-SYSTEM MAKE:DEFSYSTEM
> MAKE:EDIT-SYSTEM MAKE:DEFINED-SYSTEMS MAKE:DESCRIBE-SYSTEM
> MAKE:*COMPILE-DURING-LOAD* MAKE:FILES-WHICH-NEED-COMPILATION
> MAKE:SOFTWARE-TYPE-TRANSLATION MAKE:*FILES-MISSING-IS-AN-ERROR*
> MAKE:*DONT-REDEFINE-REQUIRE* MAKE:*BINARY-PATHNAME-DEFAULT*
> MAKE:*BIN-SUBDIR* MAKE:OPERATE-ON-SYSTEM MAKE:FIND-SYSTEM
> MAKE:*RELOAD-SYSTEMS-FROM-DISK* MAKE:*MINIMAL-LOAD* MAKE:OOS
> MAKE:AFS-BINARY-DIRECTORY MAKE:DEFINE-LANGUAGE MAKE:HARDCOPY-SYSTEM
> MAKE:LOAD-SYSTEM MAKE:COMPILER-TYPE-TRANSLATION MAKE:CLEAN-SYSTEM
> MAKE:COMPILE-SYSTEM MAKE:MACHINE-TYPE-TRANSLATION MAKE:MAKE-SYSTEM-TAG-TABLE)
> T
> * (load "start.lisp")
> ; Loading #p"/var/tmp/matlisp/start.lisp".
> [GC threshold exceeded with 2,850,824 bytes in use. Commencing GC.]
> [GC completed with 761,400 bytes retained and 2,089,424 bytes freed.]
> [GC will next occur when at least 2,761,400 bytes are in use.]
> ;; Loading #p"/var/tmp/matlisp/system.dcl".
>
> Warning: This function is undefined:
> DEFLOGICALPATH
If you're getting this warning you don't have 1.0b of matlisp, you have
1.0a.
>
> Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER:
> #() is not of type (OR CONS
> BASE-STRING
> (MEMBER NIL :UNSPECIFIC :WILD)
> COMMON-LISP::PATTERN)
>
> Restarts:
> 0: [CONTINUE] Return NIL from load of "system.dcl".
> 1: Return NIL from load of "start.lisp".
> 2: [ABORT ] Return to Top-Level.
>
> Debug (type H for help)
>
> ("DEFUN MAKE-PATHNAME" 469762728 8)[:OPTIONAL]
> Source: Error finding source:
> Error in function DEBUG::GET-FILE-TOP-LEVEL-FORM: Source file no longer exists:
> target:code/pathname.lisp.
> 0] q
>
> i am finding cmucl to be hard to work with regarding finding where
> errors occur in a large source file. do you have any hints as to how
> to jump to error using ilisp?
>
> i am not sure about the kernel::complex-double-float.
> how do i tell if this exists?
>
> thanks in advance.
>
> --
> J o h a n K u l l s t a m
> [········@ne.mediaone.net]
> edlhp
FYI:
on my linux I use:
CMU Common Lisp 18-Jan-2000, running on wow.eecs.berkeley.edu
Send questions and bug reports to your local CMU CL maintainer, or to
··········@cons.org. and ·········@cons.org. respectively.
Loaded subsystems:
Python 1.0, target Intel x86
CLOS based on PCL version: September 16 92 PCL (f)
MATLISP/1.0b
*
Johan Kullstam wrote:
>
> Raymond Toy <···@rtp.ericsson.se> writes:
>
> > >>>>> "Johan" == Johan Kullstam <········@ne.mediaone.net> writes:
> >
> >
> > Johan> hmm. i've been trying to get matlisp up and running on hp-ux with
> > Johan> cmucl 18a. it craps out in the loading of start.lisp. i get some
> > Johan> errors about logical paths. i am not at work, so i can't paste the
> >
> > Logical pathname support in CMUCL got much better in later versions of
> > CMUCL. I don't remember if 18a is a "later version".
> >
> > Johan> exact message. however, i will do so tomorrow.
> >
> > Cool.
>
> ok you asked for it ;-)
>
> from the start you can see it's 18a.
>
> CMU Common Lisp 18a, running on edlhp221
> Send bug reports and questions to your local CMU CL maintainer,
> or to ·······@mikemac.com,
> or to ··········@cons.org. (prefered)
>
> Loaded subsystems:
> Python 1.0, target HP-PA 1.1
> CLOS based on PCL version: September 16 92 PCL (f)
>
> i load up defsystem
>
> * (load "defsystem")
> ; Loading #p"/var/tmp/matlisp/defsystem.hpf".
> Warning: MAKE also exports the following symbols:
> (MAKE:UNDEFSYSTEM MAKE:SYSTEM-SOURCE-SIZE MAKE:*MULTIPLE-LISP-SUPPORT*
> MAKE:*CENTRAL-REGISTRY* MAKE:ALLEGRO-MAKE-SYSTEM-FASL
> MAKE:*DEFSYSTEM-VERSION* MAKE:AFS-SOURCE-DIRECTORY
> MAKE:*SOURCE-PATHNAME-DEFAULT* MAKE:FILES-IN-SYSTEM MAKE:DEFSYSTEM
> MAKE:EDIT-SYSTEM MAKE:DEFINED-SYSTEMS MAKE:DESCRIBE-SYSTEM
> MAKE:*COMPILE-DURING-LOAD* MAKE:FILES-WHICH-NEED-COMPILATION
> MAKE:SOFTWARE-TYPE-TRANSLATION MAKE:*FILES-MISSING-IS-AN-ERROR*
> MAKE:*DONT-REDEFINE-REQUIRE* MAKE:*BINARY-PATHNAME-DEFAULT*
> MAKE:*BIN-SUBDIR* MAKE:OPERATE-ON-SYSTEM MAKE:FIND-SYSTEM
> MAKE:*RELOAD-SYSTEMS-FROM-DISK* MAKE:*MINIMAL-LOAD* MAKE:OOS
> MAKE:AFS-BINARY-DIRECTORY MAKE:DEFINE-LANGUAGE MAKE:HARDCOPY-SYSTEM
> MAKE:LOAD-SYSTEM MAKE:COMPILER-TYPE-TRANSLATION MAKE:CLEAN-SYSTEM
> MAKE:COMPILE-SYSTEM MAKE:MACHINE-TYPE-TRANSLATION MAKE:MAKE-SYSTEM-TAG-TABLE)
> T
> * (load "start.lisp")
> ; Loading #p"/var/tmp/matlisp/start.lisp".
> [GC threshold exceeded with 2,850,824 bytes in use. Commencing GC.]
> [GC completed with 761,400 bytes retained and 2,089,424 bytes freed.]
> [GC will next occur when at least 2,761,400 bytes are in use.]
> ;; Loading #p"/var/tmp/matlisp/system.dcl".
>
> Warning: This function is undefined:
> DEFLOGICALPATH
>
> Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER:
> #() is not of type (OR CONS
> BASE-STRING
> (MEMBER NIL :UNSPECIFIC :WILD)
> COMMON-LISP::PATTERN)
>
> Restarts:
> 0: [CONTINUE] Return NIL from load of "system.dcl".
> 1: Return NIL from load of "start.lisp".
> 2: [ABORT ] Return to Top-Level.
>
> Debug (type H for help)
>
> ("DEFUN MAKE-PATHNAME" 469762728 8)[:OPTIONAL]
> Source: Error finding source:
> Error in function DEBUG::GET-FILE-TOP-LEVEL-FORM: Source file no longer exists:
> target:code/pathname.lisp.
> 0] q
>
> i am finding cmucl to be hard to work with regarding finding where
> errors occur in a large source file. do you have any hints as to how
> to jump to error using ilisp?
>
> i am not sure about the kernel::complex-double-float.
> how do i tell if this exists?
>
> thanks in advance.
>
> --
> J o h a n K u l l s t a m
> [········@ne.mediaone.net]
> edlhp