Steel Bank Common Lisp

Lars Rune N=C3=B8stdal <larsnostdal@...> writes:

> Hi,
> Ok, I have two PC's .. regular 32-bit thingies - I'm not sure what inform=
ation I
> should give - but here is the code that triggers it on both machines and =
some
> other information:

Can you post the transcript of the entire sbcl session, as well as the
contents of your /etc/sbclrc and ~/.sbclrc files?  Do you compile and
load your lisp forms, or just type them one-by-one into the repl?  We
need to see as exact a transcript as possible to try to diagnose this
problem.

(If either of your init files is particularly complex, try removing as
much as possible while the problem still exists).

Cheers,

Christophe

Christophe Rhodes <csr21 <at> cam.ac.uk> writes:

> Can you post the transcript of the entire sbcl session, as well as the
> contents of your /etc/sbclrc and ~/.sbclrc files?  Do you compile and
> load your lisp forms, or just type them one-by-one into the repl?  We
> need to see as exact a transcript as possible to try to diagnose this
> problem.


Sorry, I forgot about my .sbclrc-file. 

Ok, I think I've figured out what triggers this. When I have this in my
.sbclrc-file, it triggers the condition:

(proclaim '(optimize 
            (compilation-speed 0)
            (debug 3)
            (safety 3)
            (space 0)
            (speed 0)))


When I set either debug or safety to 0 it does not happen. This does not happen
with 0.8.20, only 0.8.21.

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

Lars Rune N=C3=B8stdal <larsnostdal@...> writes:

>             (debug 3)

This is likely the culprit.  I believe that HEAD sbcl has this bug
fixed; if you are in a position to, could you check out the CVS HEAD
and test your program?

Thanks,

Christophe

Christophe Rhodes <csr21 <at> cam.ac.uk> writes:

> 
> Lars Rune Nøstdal <larsnostdal <at> gmail.com> writes:
> 
> >             (debug 3)
> 
> This is likely the culprit.  I believe that HEAD sbcl has this bug
> fixed; if you are in a position to, could you check out the CVS HEAD
> and test your program?

Yes, I'll test with HEAD tomorrow.

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

On Sun, 17 Apr 2005 22:58:22 +0000, Lars Rune Nøstdal wrote:

> Christophe Rhodes <csr21 <at> cam.ac.uk> writes:
> 
>> 
>> Lars Rune Nøstdal <larsnostdal <at> gmail.com> writes:
>> 
>> >             (debug 3)
>> 
>> This is likely the culprit.  I believe that HEAD sbcl has this bug
>> fixed; if you are in a position to, could you check out the CVS HEAD
>> and test your program?
> 
> Yes, I'll test with HEAD tomorrow.

I'm having some trouble compiling HEAD:

cvs -z3 -d:pserver:anonymous@... co -P sbcl
...
cd sbcl
sh make.sh

..gives lot's of output, then:

; file: /home/lars/lisp/sbcl/src/compiler/globaldb.lisp
; in: define-info-type :class
;     (SB!KERNEL:CLASSOID-LAYOUT CLASS)
;
; caught style-warning:
;   undefined function: classoid-layout

;     (SB!KERNEL:FIND-CLASSOID SB!C::NAME NIL)
;
; caught style-warning:
;   undefined function: find-classoid

;     (SB!KERNEL:SPECIFIER-TYPE 'FUNCTION)
;
; caught style-warning:
;   undefined function: specifier-type

; in: defun do-compact-info
;     (SB!IMPL::BACKQ-LIST 'BLOCK
;                        SB!C::|punt|
;                        (SB!IMPL::BACKQ-LIST 'LET
;                                             (SB!IMPL::BACKQ-LIST
;                                              (SB!IMPL::BACKQ-LIST
;                                               SB!C::NAME-VAR
;                                               (SB!IMPL::BACKQ-LIST 'SVREF
;                                                                    SB!C::N-TABLE
;                                                                    SB!C::N-INDEX)))
;                                             (SB!IMPL::BACKQ-LIST 'UNLESS
;                                                                  (SB!IMPL::BACKQ-LIST*
;                                                                   'EQL
;                                                                   SB!C::NAME-VAR
;                                                                   '(0))
;                                                                  (SB!IMPL::BACKQ-LIST
;                                                                   'SB!INT:DO-ANONYMOUS
;                                                                   (SB!IMPL::BACKQ-LIST
;                                                                    #)
;                                                                   '(NIL)
;                                                                   (SB!IMPL::BACKQ-LIST
;                                                                    'DECLARE
;                                                                    #)
;                                                                   (SB!INT:ONCE-ONLY
;                                                                    #
;                                                                    #)))))
;
; caught warning:
;   undefined variable: PUNT

; in: lambda nil
;     (SB!IMPL::BACKQ-LIST 'BLOCK
;                        SB!C::|punt|
;                        (SB!IMPL::BACKQ-LIST 'LET
;                                             (SB!IMPL::BACKQ-LIST
;                                              (SB!IMPL::BACKQ-LIST
;                                               SB!C::NAME-VAR
;                                               (SB!IMPL::BACKQ-LIST 'SVREF
;                                                                    SB!C::N-TABLE
;                                                                    SB!C::N-INDEX)))
;                                             (SB!IMPL::BACKQ-LIST 'UNLESS
;                                                                  (SB!IMPL::BACKQ-LIST*
;                                                                   'EQL
;                                                                   SB!C::NAME-VAR
;                                                                   '(0))
;                                                                  (SB!IMPL::BACKQ-LIST
;                                                                   'SB!INT:DO-ANONYMOUS
;                                                                   (SB!IMPL::BACKQ-LIST
;                                                                    #)
;                                                                   '(NIL)
;                                                                   (SB!IMPL::BACKQ-LIST
;                                                                    'DECLARE
;                                                                    #)
;                                                                   (SB!INT:ONCE-ONLY
;                                                                    #
;                                                                    #)))))
;
; caught warning:
;   undefined variable: PUNT

;
; caught warning:
;   This variable is undefined:
;     PUNT

;
; caught style-warning:
;   These functions are undefined:
;     classoid-layout find-classoid specifier-type
;
; compilation unit finished
;   caught 1 ERROR condition
;   caught 3 WARNING conditions
;   caught 6 STYLE-WARNING conditions
;   printed 39 notes

; /home/lars/lisp/sbcl/obj/from-host/src/compiler/globaldb.lisp-obj-tmp written
; compilation finished in 0:00:05
unhandled simple-error in thread 29532:
  FAILURE-P was set when creating "obj/from-host/src/compiler/globaldb.lisp-obj".

0: (sb-debug:backtrace 128 #<synonym-stream :SYMBOL sb-sys:*stderr* {5066859}>)
1: (sb-debug::debugger-disabled-hook
    #<simple-error {95B94E9}>
    #<unavailable argument>)
2: (invoke-debugger #<simple-error {95B94E9}>)
3: (error #<simple-error {95B94E9}>)
4: (compile-stem
    "src/compiler/globaldb"
    :obj-prefix
    "obj/from-host/"
    :obj-suffix
    ".lisp-obj"
    :tmp-obj-suffix-suffix
    "-tmp"
    :src-prefix
    ""
    :src-suffix
    ".lisp"
    :compile-file
    #<FUNCTION compile-file>
    :trace-file
    nil
    :ignore-failure-p
    nil)
5: ((lambda ()))
6: (in-host-compilation-mode #<CLOSURE (lambda #) {B660C05}>)
7: (host-cload-stem "src/compiler/globaldb" :ignore-failure-p nil)
8: (load-or-cload-xcompiler #<FUNCTION host-cload-stem>)
9: (sb-int:eval-in-lexenv
    (load-or-cload-xcompiler #'host-cload-stem)
    #S(sb-kernel:lexenv
       :funs nil
       :vars nil
       :blocks nil
       :tags nil
       :type-restrictions nil
       ...))
10: (sb-ext:interactive-eval (load-or-cload-xcompiler #'host-cload-stem))
11: (sb-impl::repl-fun nil)
12: ((lambda ()))
13: ((lambda ()))
14: (sb-impl::%with-rebound-io-syntax #<CLOSURE (lambda #) {906DE6D}>)
15: (sb-impl::toplevel-repl nil)
16: (sb-impl::toplevel-init)
17: ((flet sb-impl::restart-lisp))

unhandled condition in --disable-debugger mode, quitting
deleted #P"/home/lars/lisp/sbcl/obj/from-host/src/compiler/globaldb.lisp-obj-tmp"

real    0m58.676s
user    0m46.740s
sys     0m2.131s


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

Lars Rune N=F8stdal <larsnostdal@...> writes:

> cd sbcl
> sh make.sh
>
> ..gives lot's of output, then:
             ^-- argh! :-)

> ; in: defun do-compact-info
> ;     (SB!IMPL::BACKQ-LIST 'BLOCK
> ;                        SB!C::|punt|
> ; caught warning:
> ;   undefined variable: PUNT

Oh, but this is interesting!  I don't know if anyone's ever tried to
compile sbcl with readtable-case :invert before.

The short answer might be to do

  sh make.sh 'sbcl --userinit /dev/null'

but if you want to keep your :invert, try changing the PUNTs in
do-compact-info (in src/compiler/globaldb) to lowercase.

Cheers,

Christophe

On Tue, 19 Apr 2005 09:22:38 +0100, Christophe Rhodes wrote:

> Lars Rune Nøstdal <larsnostdal@...> writes:
> 
>> cd sbcl
>> sh make.sh
>>
>> ..gives lot's of output, then:
>              ^-- argh! :-)
> 
>> ; in: defun do-compact-info
>> ;     (SB!IMPL::BACKQ-LIST 'BLOCK
>> ;                        SB!C::|punt|
>> ; caught warning:
>> ;   undefined variable: PUNT
> 
> Oh, but this is interesting!  I don't know if anyone's ever tried to
> compile sbcl with readtable-case :invert before.

err .. hehe :)
 
> The short answer might be to do
> 
>   sh make.sh 'sbcl --userinit /dev/null'
> 
> but if you want to keep your :invert, try changing the PUNTs in
> do-compact-info (in src/compiler/globaldb) to lowercase.

i'll try both - just for fun .. :p
i wonder .. what happens when i afterwards use :preserve in my brand new
SBCL?

(SBCL-compiles takes forever here btw. ..)

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

On Tue, 19 Apr 2005 13:11:23 +0200, Lars Rune Nøstdal wrote:

> On Tue, 19 Apr 2005 09:22:38 +0100, Christophe Rhodes wrote:
> 
>> Lars Rune Nøstdal <larsnostdal@...> writes:
>> 
>>> cd sbcl
>>> sh make.sh
>>>
>>> ..gives lot's of output, then:
>>              ^-- argh! :-)
>> 
>>> ; in: defun do-compact-info
>>> ;     (SB!IMPL::BACKQ-LIST 'BLOCK
>>> ;                        SB!C::|punt|
>>> ; caught warning:
>>> ;   undefined variable: PUNT
>> 
>> Oh, but this is interesting!  I don't know if anyone's ever tried to
>> compile sbcl with readtable-case :invert before.
> 
> err .. hehe :)
>  
>> The short answer might be to do
>> 
>>   sh make.sh 'sbcl --userinit /dev/null'
>> 
>> but if you want to keep your :invert, try changing the PUNTs in
>> do-compact-info (in src/compiler/globaldb) to lowercase.
> 
> i'll try both - just for fun .. :p
> i wonder .. what happens when i afterwards use :preserve in my brand new
> SBCL?
> 
> (SBCL-compiles takes forever here btw. ..)

ok - it seems to be fixed in HEAD as you said

the other experiment (compiling with :invert) failed in the end after
trying to convert many of THESE into these .. i'll try it again later i
think - (and/or maybe compiling with :preserve)?

i'm working on some other stuff now, and tuning down the
debug/safety-settings works - so i'll just do that until a new release is
out

thanks .. :)

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

On Tue, 19 Apr 2005, [iso-8859-1] Lars Rune N=F8stdal wrote:

> the other experiment (compiling with :invert) failed in the end after
> trying to convert many of THESE into these .. i'll try it again later i
> think - (and/or maybe compiling with :preserve)?

May I ask what's the point of this excercise? What's the benefit of being=
=20
able to build SBCL with non-upcasing readtable?

Of course, the easist way to make SBCL buildable so would be to have SBCL=
=20
fix the readtable for building to a known-good state -- but this is=20
obviously not what you're after.

Cheers,

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

On Tue, 19 Apr 2005 22:57:17 +0300, Nikodemus Siivola wrote:

> On Tue, 19 Apr 2005, [iso-8859-1] Lars Rune Nøstdal wrote:
> 
>> the other experiment (compiling with :invert) failed in the end after
>> trying to convert many of THESE into these .. i'll try it again later i
>> think - (and/or maybe compiling with :preserve)?
> 
> May I ask what's the point of this excercise? What's the benefit of being 
> able to build SBCL with non-upcasing readtable?

For fun, and maybe if I built it with :downcase for example, I would be
able to use :preserve and not have to use upper-case names for stuff built
during complation of SBCL itself ..

> Of course, the easist way to make SBCL buildable so would be to have SBCL 
> fix the readtable for building to a known-good state -- but this is 
> obviously not what you're after.

not sure I understand

> Cheers,
> 
>   -- Nikodemus              Schemer: "Buddha is small, clean, and serious."
>                    Lispnik: "Buddha is big, has hairy armpits, and laughs."
-- 
mvh,
Lars Rune Nøstdal
http://lars.nostdal.org/

Lars Rune N=C3=B8stdal <larsnostdal@...> writes:

> On Tue, 19 Apr 2005 22:57:17 +0300, Nikodemus Siivola wrote:
>
>> May I ask what's the point of this excercise? What's the benefit of bein=
g=20
>> able to build SBCL with non-upcasing readtable?
>
> For fun,=20

That's a valid reason.  (I'm trying it myself in an attempt to
harmonize notation; a non-case-folding build enforces a consistent
capitalization of like names).

> and maybe if I built it with :downcase for example, I would be able
> to use :preserve and not have to use upper-case names for stuff
> built during complation of SBCL itself ..

This will never, ever work.

Cheers,

Christophpe