sbcl-help — Support channel for SBCL users and bug-reports.

Re: [Sbcl-help] build error om win32

[Rahul J. <rj...@ny...>]

Nikodemus Siivola <nik...@ra...> writes:

> Rahul Jain <rj...@ny...> writes:
>
>>> gcc -g -Wall -O3 -mno-cygwin -I.  -c -o x86-assem.o x86-assem.S
>>> x86-assem.S:1089:9: warning: no newline at end of file
>>> x86-assem.S: Assembler messages:
>>> x86-assem.S:368: Warning: .type pseudo-op used outside of .def/.endef 
>>> ignored.
>
>> Looks like you've got newline issues. Make sure the file and your
>> newline settings in cygwin match up.
>
> Erm, no. 
>
> x86-assem.S really has a missing newline at the end of the file, and
> the pseudo-op warning is real. (Fixed in CVS.)

Fair enough. :)

-- 
Rahul Jain
rj...@ny...
Professional Software Developer, Amateur Quantum Mechanicist

Re: [Sbcl-help] build error om win32

[Nikodemus S. <nik...@ra...>]

Rahul Jain <rj...@ny...> writes:

>> gcc -g -Wall -O3 -mno-cygwin -I.  -c -o x86-assem.o x86-assem.S
>> x86-assem.S:1089:9: warning: no newline at end of file
>> x86-assem.S: Assembler messages:
>> x86-assem.S:368: Warning: .type pseudo-op used outside of .def/.endef 
>> ignored.

> Looks like you've got newline issues. Make sure the file and your
> newline settings in cygwin match up.

Erm, no. 

x86-assem.S really has a missing newline at the end of the file, and
the pseudo-op warning is real. (Fixed in CVS.)

Cheers,

  -- Nikodemus              Schemer: "Buddha is small, clean, and serious."
                   Lispnik: "Buddha is big, has hairy armpits, and laughs."

Re: [Sbcl-help] build error om win32

[Rahul J. <rj...@ny...>]

Yaroslav Kavenchuk <kav...@tu...> writes:

> sbcl 0.9.17.10
>
> build log:
>
> gcc -g -Wall -O3 -mno-cygwin -I.  -c -o x86-assem.o x86-assem.S
> x86-assem.S:1089:9: warning: no newline at end of file
> x86-assem.S: Assembler messages:
> x86-assem.S:368: Warning: .type pseudo-op used outside of .def/.endef 
> ignored.
[...]

Looks like you've got newline issues. Make sure the file and your
newline settings in cygwin match up.


-- 
Rahul Jain
rj...@ny...
Professional Software Developer, Amateur Quantum Mechanicist

[Sbcl-help] Windows installer for 0.9.17

[Nikodemus S. <nik...@ra...>]

I added a Windows installer to the 0.9.17 release files:

 http://prdownloads.sourceforge.net/sbcl/sbcl-0.9.17-x86-windows-binary.msi?download

Built using the make-windows-installer.sh, except I removed the Program Menu
shortcut from the installer, as it needs to be manually tweaked in order to
work: 

 If you make shortcuts to the 0.9.17 binary and core, you need to manually
 disable the "Advanced Text Services" for the shortcut from the properties /
 compatibility tab.

Cheers,

  -- Nikodemus              Schemer: "Buddha is small, clean, and serious."
                   Lispnik: "Buddha is big, has hairy armpits, and laughs."

Re: [Sbcl-help] problem compiling sbcl

[Andras S. <as...@ma...>]

Andras Simon <as...@ma...> writes:

> Juho Snellman <js...@ik...> writes:
> 
> > Andras Simon <as...@ma...> writes:
> > > (Meanwhile, I've downloaded the 0.9.17 binary from sf and it seems to
> > > run fine. Should I try to compile sbcl with it?)
> > 
> > I wouldn't expect that to make any difference. But can you run the
> 
> Compiling with sbcl worked, so I thought it did make a difference. 
> But then...
> 
> > following tests, and report which ones work?
> > 
> >   src/runtime/sbcl --core output/cold-sbcl.core --no-userinit
> > 
> >   (defun sb!impl::profile-deinit ())
> >   (save-lisp-and-die "foo.core")
> > 
> > And:
> > 
> >   src/runtime/sbcl --core foo.core --no-userinit
> > 
> > And:
> > 
> >   src/runtime/sbcl --core output/sbcl.core --no-userinit
> 
> ...I recompiled with cmucl to be able to run these tests, and this
> time it went without a hitch. (I didn't update the sources in the
> meantime.) I'll try it again, and will run the tests if there is a
> problem. 

Still no problem after several recompilations. Except that running 
sh ./run-tests.sh produces this:

Finished running tests.
Status:
 Expected failure:    debug.impure.lisp / (UNDEFINED-FUNCTION BUG-353)
 Unexpected success:  debug.impure.lisp / (THROW NO-SUCH-TAG)
 Unexpected success:  debug.impure.lisp / (BACKTRACE MISC)
 Expected failure:    external-format.impure.lisp /
(CHARACTER-DECODE-LARGE
                                                     FORCE-END-OF-FILE)
 Failure:             timer.impure.lisp / (WITH-TIMEOUT MANY-AT-THE-SAME-TIME)
test failed, expected 104 return code, got 1

I guess it's unrelated, but I thought I'd mention it. 

Andras

Re: [Sbcl-help] problem compiling sbcl

[Andras S. <as...@ma...>]

Juho Snellman <js...@ik...> writes:

> Andras Simon <as...@ma...> writes:
> > (Meanwhile, I've downloaded the 0.9.17 binary from sf and it seems to
> > run fine. Should I try to compile sbcl with it?)
> 
> I wouldn't expect that to make any difference. But can you run the

Compiling with sbcl worked, so I thought it did make a difference. 
But then...

> following tests, and report which ones work?
> 
>   src/runtime/sbcl --core output/cold-sbcl.core --no-userinit
> 
>   (defun sb!impl::profile-deinit ())
>   (save-lisp-and-die "foo.core")
> 
> And:
> 
>   src/runtime/sbcl --core foo.core --no-userinit
> 
> And:
> 
>   src/runtime/sbcl --core output/sbcl.core --no-userinit

...I recompiled with cmucl to be able to run these tests, and this
time it went without a hitch. (I didn't update the sources in the
meantime.) I'll try it again, and will run the tests if there is a
problem. 

Andras

[Sbcl-help] build error om win32

[Yaroslav K. <kav...@tu...>]

sbcl 0.9.17.10

build log:

gcc -g -Wall -O3 -mno-cygwin -I.  -c -o x86-assem.o x86-assem.S
x86-assem.S:1089:9: warning: no newline at end of file
x86-assem.S: Assembler messages:
x86-assem.S:368: Warning: .type pseudo-op used outside of .def/.endef 
ignored.
x86-assem.S:368: Error: junk at end of line, first unrecognized 
character is `_'
x86-assem.S:374: Warning: .size pseudo-op used outside of .def/.endef 
ignored.
x86-assem.S:374: Error: junk at end of line, first unrecognized 
character is `_'
make: *** [x86-assem.o] Error 1

-- 
WBR, Yaroslav Kavenchuk.

Re: [Sbcl-help] problem compiling sbcl

[Juho S. <js...@ik...>]

Andras Simon <as...@ma...> writes:
> (Meanwhile, I've downloaded the 0.9.17 binary from sf and it seems to
> run fine. Should I try to compile sbcl with it?)

I wouldn't expect that to make any difference. But can you run the
following tests, and report which ones work?

  src/runtime/sbcl --core output/cold-sbcl.core --no-userinit

  (defun sb!impl::profile-deinit ())
  (save-lisp-and-die "foo.core")

And:

  src/runtime/sbcl --core foo.core --no-userinit

And:

  src/runtime/sbcl --core output/sbcl.core --no-userinit

-- 
Juho Snellman

[Sbcl-help] Sorry, it's really urgent

[Gerardo h. <ma...@gm...>]

I apologize for using this list to distribute information which isn't
at all related to Lisp. But we need to get this information to the
rest of the world ASAP.


Copter flyovers raise Mexico tourist town tensions
http://today.reuters.co.uk/news/CrisesArticle.aspx?storyId=3DN30254192&WTmo=
dLoc=3DWorld-R5-Alertnet-3


That's not the whole story. There is more (in spanish):

Tirotean campamento de maestros en Oaxaca
http://www.eldiariony.com/noticias/detail.aspx?section=3D20&desc=3DNuestros=
%20Pa=EDses&id=3D1499920

Unknown persons are shooting at people from APPO to make them react
violently and to provide a justification to send in the PFP against
them. PFP means Policia Federal Preventiva  (Federal Preventive
Police) but don't let the name fool you, it's the army.

The army is about to enter Oaxaca. It will happen any minute now. This
is really sad.


URGENT SOLIDARITY CALL FOR OAXACA
http://www.indybay.org/newsitems/2006/09/29/18316698.php

URGENT SOLIDARITY CALL FOR OAXACA

At the federal level in Mexico, the current discourse signals an
imminent arrival of Federal Police Forces in Oaxaca. The feds claim
that, if federal forces are sent to Oaxaca, they will only maintain a
presence on the outskirts of the city, to "ensure civilian safety."
However, it is widely known that local PRI-sympathizing groups can be
mobilized to provoke a confrontation with the sectors of civil society
partiapting in the popular movement, which would justify the entrance
of the federal police.

If the federal police enter Oaxaca, it will be a blood bath...

Please call or send faxes and emails to President Fox and to Secretary
of Interior Affairs, Carlos Abascal, demanding the immediate
withdrawal of threats to send police forces into Oaxaca, and the
immediate resignation of Oaxacan governor, Ulises Ruiz Ortiz. Write in
Spanish. Write in English. Just write, or call, or both.

Get down to your local Mexican consulate or embassy. Make a lot of
noise. Spend the night out front if you have to.


President Vicente Fox:
Email: vicente.fox.quesadda [at] presidencia.gob.mx
Fax: 011-52-55-52-77-23-76
Phone: 011-52-55-27-89-11-00

Sec. of Internal Affairs, Carlos Abascal
Tel: 011-52-55-50-93-34-00
Email: cabascal [at] segob.gob.mx

--=20
Gerardo Horvilleur
ma...@si...
5813-0830

Re: [Sbcl-help] problem compiling sbcl

[Andras S. <as...@ma...>]

[I sent this a few hours ago but it didn't show up on gmane.]

Andras Simon <as...@ma...> writes:

> Christophe Rhodes <cs...@ca...> writes:
> 
> > Andras Simon <as...@ma...> writes:
> > 
> 
> >                                       and print the two things in the
> > error message ("print $1" and "print $2")
> 
> You've lost me here :( Which error message and where should I type
> "print $1"? Sorry for being so thick!

Apologies again! Here we go:

ldb> print 0x0a042f77
$1=       0x0a042f77: other pointer
            header: 0x00000000: 0
          (invalid header object)
ldb> print 0x096db57b
$2=       0x096db57b: list pointer
$3=         car: 0x09051f2f: OR
$4=         cdr: 0x096db583: ($5=($6=VECTOR $7=CHARACTER) $8=($6
$nil=NIL) $9=BASE-STRING $10=SYMBOL ...)

I hope this is what you meant.

Andras

Re: [Sbcl-help] problem compiling sbcl

[Andras S. <as...@ma...>]

Andras Simon <as...@ma...> writes:

> Christophe Rhodes <cs...@ca...> writes:
> 
> > Andras Simon <as...@ma...> writes:
> > 
> 
> >                                       and print the two things in the
> > error message ("print $1" and "print $2")
> 
> You've lost me here :( Which error message and where should I type
> "print $1"? Sorry for being so thick!

Apologies again! Here we go:

ldb> print 0x0a042f77
$1=       0x0a042f77: other pointer
            header: 0x00000000: 0
          (invalid header object)
ldb> print 0x096db57b
$2=       0x096db57b: list pointer
$3=         car: 0x09051f2f: OR
$4=         cdr: 0x096db583: ($5=($6=VECTOR $7=CHARACTER) $8=($6
$nil=NIL) $9=BASE-STRING $10=SYMBOL ...)

I hope this is what you meant.

Andras

Re: [Sbcl-help] problem compiling sbcl

[Andras S. <as...@ma...>]

Christophe Rhodes <cs...@ca...> writes:

> Andras Simon <as...@ma...> writes:
> 
> > real    2m16.228s
> > user    2m3.879s
> > sys     0m3.233s
> > make: Entering directory `/home/algebra/asimon/LISP/sbcl/contrib/asdf'
> > make: Warning: File `../../output/sbcl.core' has modification time 1.7e+02 s in the future
> >
> > Can the above warning make gives have something to do with what
> > happens? Is there any other detail I should provide to make this a
> > meaningful bug report?
> 
> I don't think that the warning is relevant.  What is your locale like,
> particularly the character set stuff?  You could also provide a

tarski:~$ locale
LANG=
LC_CTYPE="POSIX"
LC_NUMERIC="POSIX"
LC_TIME="POSIX"
LC_COLLATE="POSIX"
LC_MONETARY="POSIX"
LC_MESSAGES="POSIX"
LC_PAPER="POSIX"
LC_NAME="POSIX"
LC_ADDRESS="POSIX"
LC_TELEPHONE="POSIX"
LC_MEASUREMENT="POSIX"
LC_IDENTIFICATION="POSIX"
LC_ALL=
tarski:~$ 


> backtrace from ldb (type "backtrace") 

See my followup to Juho's post.

>                                       and print the two things in the
> error message ("print $1" and "print $2")

You've lost me here :( Which error message and where should I type
"print $1"? Sorry for being so thick!

Andras

Re: [Sbcl-help] problem compiling sbcl

[Andras S. <as...@ma...>]

Juho Snellman <js...@ik...> writes:

> A backtrace from here might provide some hint about out what's wrong
> (type "back" in the ldb prompt).

Here it is: 

ldb> back
Backtrace:
   0: Foreign function (null), fp = 0xb7c2410c, ra = 0x805658b
   1: Foreign function ldb_monitor, fp = 0xb7c241dc, ra = 0x80560fe
   2: Foreign function lose, fp = 0xb7c241fc, ra = 0x805318e
   3: Foreign function (null), fp = 0xb7c2421c, ra = 0x8054696
   4: Foreign function sigtrap_handler, fp = 0xb7c2424c, ra =
0x805dadf
   5: Foreign function (null), fp = 0xb7c242ec, ra = 0x8055aca
   6: Foreign function __kernel_rt_sigreturn, fp = 0xb7c246d0, ra =
0xffffe440
   7: Foreign fp = 0xb7c246f0, ra = 0x9cc5ceb
fatal error encountered in SBCL pid 9353(tid 3085355328):
no size function for object at 0x09cc4228 (widetag 0x28)

   8: LDB monitor
ldb> 

(Meanwhile, I've downloaded the 0.9.17 binary from sf and it seems to
run fine. Should I try to compile sbcl with it?)

Andras

Re: [Sbcl-help] problem compiling sbcl

[Christophe R. <cs...@ca...>]

Andras Simon <as...@ma...> writes:

> real    2m16.228s
> user    2m3.879s
> sys     0m3.233s
> make: Entering directory `/home/algebra/asimon/LISP/sbcl/contrib/asdf'
> make: Warning: File `../../output/sbcl.core' has modification time 1.7e+02 s in the future
>
> Can the above warning make gives have something to do with what
> happens? Is there any other detail I should provide to make this a
> meaningful bug report?

I don't think that the warning is relevant.  What is your locale like,
particularly the character set stuff?  You could also provide a
backtrace from ldb (type "backtrace") and print the two things in the
error message ("print $1" and "print $2")

Cheers,

Christophe

Re: [Sbcl-help] problem compiling sbcl

[Juho S. <js...@ik...>]

Andras Simon <as...@ma...> writes:
> make: Entering directory `/home/algebra/asimon/LISP/sbcl/contrib/asdf'
> make: Warning: File `../../output/sbcl.core' has modification time 1.7e+02 s in the future
> /home/algebra/asimon/LISP/sbcl/src/runtime/sbcl --noinform --core
> /home/algebra/asimon/LISP/sbcl/output/sbcl.core --disable-debugger
> --no-sysinit --no-userinit --eval '(compile-file (format nil
> "SYS:CONTRIB;~:@(~A~);~:@(~A~).LISP" "asdf" "asdf"))' </dev/null
> internal error #31
>     SC: 14, Offset: 0   $1=       0x0a042f77: other pointer
>     SC: 14, Offset: 2   $2=       0x096db57b: list pointer
> fatal error encountered in SBCL pid 9353(tid 3085355328):
> internal error too early in init, can't recover
> 
> LDB monitor
> ldb>  

A backtrace from here might provide some hint about out what's wrong
(type "back" in the ldb prompt).

-- 
Juho Snellman

[Sbcl-help] problem compiling sbcl

[Andras S. <as...@ma...>]

While trying to compile cvs sbcl with cmucl (October snapshot), 
I got

* [undoing binding stack and other enclosing state... done]
[saving current Lisp image into
/home/algebra/asimon/LISP/sbcl/output/sbcl.core:
writing 1912 bytes from the read-only space at 0x01000000
writing 1936 bytes from the static space at 0x05000000
writing 23613440 bytes from the dynamic space at 0x09000000
done]

real    2m16.228s
user    2m3.879s
sys     0m3.233s
make: Entering directory `/home/algebra/asimon/LISP/sbcl/contrib/asdf'
make: Warning: File `../../output/sbcl.core' has modification time 1.7e+02 s in the future
/home/algebra/asimon/LISP/sbcl/src/runtime/sbcl --noinform --core
/home/algebra/asimon/LISP/sbcl/output/sbcl.core --disable-debugger
--no-sysinit --no-userinit --eval '(compile-file (format nil
"SYS:CONTRIB;~:@(~A~);~:@(~A~).LISP" "asdf" "asdf"))' </dev/null
internal error #31
    SC: 14, Offset: 0   $1=       0x0a042f77: other pointer
    SC: 14, Offset: 2   $2=       0x096db57b: list pointer
fatal error encountered in SBCL pid 9353(tid 3085355328):
internal error too early in init, can't recover

LDB monitor
ldb>  

customize-target-features.lisp is 

(lambda (list)
  (flet ((enable (x) (pushnew x list))
         (disable (x) (setf list (remove x list))))
    (enable :sb-thread)
;    (enable :sb-futex)
    #+nil (enable :sb-show)
    list))

and the box this happens on is

tarski:~$ uname -a
Linux tarski 2.6.10-ac10 #1 Thu Jan 20 11:56:48 CET 2005 i686
GNU/Linux

Can the above warning make gives have something to do with what
happens? Is there any other detail I should provide to make this a
meaningful bug report?

Andras

[Sbcl-help] PHogaARMA

[Socorro C. <mi...@ca...>]

Good day,

VALttIUM

AMBttIEN

CIAttLIS

VIAttGRA

Save 50 % with http://www.fendrionkadefunher.com
=20
  _____ =20


My cat can do better. And scratch dirt over it.
television. I am sure that it will cheer you to know that, since your
Now that is a pleasure to hear, visiting gentlemen. Before my

Re: [Sbcl-help] Symbol weirdness

[<lar...@gm...>]

On Mon, 02 Oct 2006 03:05:24 +0200, Lars Rune Nøstdal wrote:

> ;; this is set in ~/.sbclrc
> ;; (setf (readtable-case *readtable*) :invert)
> 
> 
> (defun what-is-going-on? (sym)
>   (let ((str (string sym)))
>     (setf (char str 0)
>           (char-downcase (char str 0)))))
> 
> 
> (defun test ()
>   (let ((sym 'Person))
>          (format t "before: ~A~%" sym)
>          (format t "middle: ~A~%" (what-is-going-on? sym))
>          (format t "after: ~A (why is `sym' changed?)~%" sym)))
> 
> 
> #|
> cl-user> (test)
> before: Person
> middle: p
> after: PERSON (why is `sym' changed?)
> nil
> cl-user> (test)
> before: PERSON
> middle: p
> after: PERSON (why is `sym' changed?)
> nil
> cl-user> (test)
> before: PERSON
> middle: p
> after: PERSON (why is `sym' changed?)
> nil
> 
> ;; If I reevaluate the buffer with the code in it (C-c-k), then start
> ;; calling test again I get the same result:
> 
> cl-user> (test)
> before: Person
> middle: p
> after: PERSON (why is `sym' changed?)
> nil
> cl-user> (test)
> before: PERSON
> middle: p
> after: PERSON (why is `sym' changed?)
> nil
> cl-user> (test)
> before: PERSON
> middle: p
> after: PERSON (why is `sym' changed?)
> nil
> cl-user> 
> |#
> 
> I'm lost - what's going on here?

Ok, well the guys at #lisp figured this one out rather quickly:

(defun what-is-going-on? (sym)
  (let ((str (copy-seq (string sym)))) ;; NOTE: `copy-seq'
    (setf (char str 0)
          (char-downcase (char str 0)))))

Thnx guys .. :)

-- 
Lars Rune Nøstdal
http://lars.nostdal.org/

[Sbcl-help] Symbol weirdness

[<lar...@gm...>]

;; this is set in ~/.sbclrc
;; (setf (readtable-case *readtable*) :invert)


(defun what-is-going-on? (sym)
  (let ((str (string sym)))
    (setf (char str 0)
          (char-downcase (char str 0)))))


(defun test ()
  (let ((sym 'Person))
         (format t "before: ~A~%" sym)
         (format t "middle: ~A~%" (what-is-going-on? sym))
         (format t "after: ~A (why is `sym' changed?)~%" sym)))


#|
cl-user> (test)
before: Person
middle: p
after: PERSON (why is `sym' changed?)
nil
cl-user> (test)
before: PERSON
middle: p
after: PERSON (why is `sym' changed?)
nil
cl-user> (test)
before: PERSON
middle: p
after: PERSON (why is `sym' changed?)
nil

;; If I reevaluate the buffer with the code in it (C-c-k), then start
;; calling test again I get the same result:

cl-user> (test)
before: Person
middle: p
after: PERSON (why is `sym' changed?)
nil
cl-user> (test)
before: PERSON
middle: p
after: PERSON (why is `sym' changed?)
nil
cl-user> (test)
before: PERSON
middle: p
after: PERSON (why is `sym' changed?)
nil
cl-user> 
|#

I'm lost - what's going on here?

-- 
Lars Rune Nøstdal
http://lars.nostdal.org/

Re: [Sbcl-help] External formats and line endings

[Christophe R. <cs...@ca...>]

Edi Weitz <ed...@ag...> writes:

> Does SBCL have a facility to describe how line endings are treated,
> like, say, '(:LATIN-1 :EOL-STYLE :LF) in LispWorks?  My apologies if
> this is in the documentation somewhere, but sbcl.org is down for me.

Not at the moment.  (It's an obvious missing piece of our external
format support, along with byte-order-mark and possibly replacement
character treatment.)  At present, all external formats implicitly
treat the end-of-line as linefeed only (with the pain that this causes
on Windows) and does not treat an initial #xfeff as a byte-order-mark.

Fixing this would be nice.  It might not even be too hard now... a
nice gentle project for someone to tackle in their spare time ;-)

Cheers,

Christophe

Re: [Sbcl-help] External formats and line endings

[Edi W. <ed...@ag...>]

On Sun, 1 Oct 2006 07:13:13 -0400, Kevin Reid <kp...@ma...> wrote:

>> My apologies if this is in the documentation somewhere, but
>> sbcl.org is down for me.
>
> You can access the site as <http://sbcl.sourceforge.net/>.

Yeah, I tried that last night, but the "Manual" link leads back to
sbcl.org... :(

Re: [Sbcl-help] External formats and line endings

[Kevin R. <kp...@ma...>]

On Oct 1, 2006, at 6:45, Edi Weitz wrote:

> Does SBCL have a facility to describe how line endings are treated,  
> like, say, '(:LATIN-1 :EOL-STYLE :LF) in LispWorks?

No.

> My apologies if this is in the documentation somewhere, but  
> sbcl.org is down for me.

You can access the site as <http://sbcl.sourceforge.net/>.

-- 
Kevin Reid                            <http://homepage.mac.com/kpreid/>

[Sbcl-help] External formats and line endings

[Edi W. <ed...@ag...>]

Does SBCL have a facility to describe how line endings are treated,
like, say, '(:LATIN-1 :EOL-STYLE :LF) in LispWorks?  My apologies if
this is in the documentation somewhere, but sbcl.org is down for me.

Thanks,
Edi.

Re: [Sbcl-help] GC bug on win32

[Nikodemus S. <nik...@ra...>]

"Michael Ben Yosef" <sep...@mw...> writes:

> Below this message follows a copy of my session. While copy-pasting to sbcl,
> the "deftype"'s and the "defparameter" are read-eval-printed fine, but when
> the "defun" is entered at the prompt ... Well, see for yourself: Sbcl
> crashes and I get dropped into ldb.
>
> There's no problem if I enter the "defun" first followed by the other
> def*'s.
>
> Sorry if this is a bit of a useless bug report. I imagine it must be tough
> getting to the bottom of these GC issues. And sorry too if sbcl on win32 is
> not yet stable enough for this kind of thing to be worth reporting anyway.

While this is a known bug on Win32, it still hasn't made its way to
BUGS yet, so it is definitely worth reporting. And it is always good
have more test-cases.

As a slightly unrelated piece of good news, I've finally made some headway
in the "SBCL started from a shortcut utterly broken": it seems that turning of the
"advanced text services" for the shortcut fixes this. Still working on the
why and how, though.

Cheers,

  -- Nikodemus              Schemer: "Buddha is small, clean, and serious."
                   Lispnik: "Buddha is big, has hairy armpits, and laughs."

[Sbcl-help] GC bug on win32

[Michael B. Y. <sep...@mw...>]

Hi all,

Upon reading the recent c.l.l thread entitled "Different behavior of deftype
on clisp and sbcl. How it should really work?", I tried out the original
poster's code in SBCL 0.9.17 on Windows XP.

Below this message follows a copy of my session. While copy-pasting to sbcl,
the "deftype"'s and the "defparameter" are read-eval-printed fine, but when
the "defun" is entered at the prompt ... Well, see for yourself: Sbcl
crashes and I get dropped into ldb.

There's no problem if I enter the "defun" first followed by the other
def*'s.

Sorry if this is a bit of a useless bug report. I imagine it must be tough
getting to the bottom of these GC issues. And sorry too if sbcl on win32 is
not yet stable enough for this kind of thing to be worth reporting anyway.

Regards,

Michael



This is SBCL 0.9.17, an implementation of ANSI Common Lisp.
More information about SBCL is available at <http://www.sbcl.org/>.

SBCL is free software, provided as is, with absolutely no warranty.
It is mostly in the public domain; some portions are provided under
BSD-style licenses.  See the CREDITS and COPYING files in the
distribution for more information.

This is experimental prerelease support for the Windows platform: use
at your own risk.  "Your Kitten of Death awaits!"
* (deftype whitespace () '(member #\Space #\Newline #\Return #\Tab))

WHITESPACE
* (deftype digit () '(member #\0 #\1 #\2 #\3 #\4 #\5 #\6 #\7 #\8 #\9))

DIGIT
* (deftype punctuation () '(member #\. #\, #\! #\? #\; #\" #\' #\] #\[ #\(
#\) #
\\ #\/  #\{ #\} #\:))

PUNCTUATION
* (deftype text-type () '(not (or digit whitespace punctuation)))

TEXT-TYPE
*
(defparameter test-string
"abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ.
")

TEST-STRING
*
(defun test-text-type ()
  (loop
     for i from 0 to (1- (length test-string))
     collect (typep (aref test-string i) 'text-type)))
fatal error encountered in SBCL pid 3028:
GC invariant lost, file "gencgc.c", line 831

LDB monitor
ldb>