Re: [Sbcl-devel] Re: bug with ppc sbcl and iterate

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

On Feb 18, 2004, at 1:43 AM, Andreas Fuchs wrote:

> Today, Robert Sedgewick <rd...@me...> wrote:
>> Has anyone had any luck using the iterate package with sbcl on
>> darwin?
>
> Never tried that before...
>
>> With iterate 1.0.2 and sbcl 0.8.7.52 (and also 0.8.5.something) I
>> get a "TYPE-ERROR in thread 914: The value 2 is not of type LIST."
>> when I """
>> * (in-package :iterate)
>> #<PACKAGE "ITERATE">
>> * (iter (for i from 0 below 10)
>> 	  (collect (* 4 i)))
>> """
>>
>> This code works in openmcl and in sbcl on x86.
>
> I saw it work on sbcl/x86, too. Can you post a few macroexpand-1 steps
> of:
>
> * (macroexpand-1 '(iter (for i from 0 below 10)
>    	                (collect (* 4 i))))
> * (macroexpand-1 (macroexpand-1 '(iter (for i from 0 below 10)
>    	                               (collect (* 4 i)))))
>

I should have mentioned that the error is during macroexpansion.

> , etc on sbcl/darwin? That would help a bit, as I don't have access to
> sbcl on a darwin machine at the moment.

Well, I would give you an account here, but all I have is a laptop that 
is behind a firewall.....


>> Poking around a little it looks like something screwy is going on
>> with the multiple-value-bind in the function walk-list-nconcing in
>> iterate.lisp.  Something which should be nil gets set to something
>> random (in this case 2).
>
> Hmm. So is this condition raised during macroexpansion or when the
> resulting code is executed? As I said above, a few MACROEXPAND-1s
> would help. Also, you could try (setf *BREAK-ON-SIGNALS* t) and
> extract a backtrace.

Setting *break-on-signals* doesn't change the error or where it occurs. 
The problem is that, in walk-list-nconcing, augment (which expands to 
nconc) gets called with a scalar, which, of course, isn't right.  How 
it is getting a scalar is a mystery to me since the form that 
multiple-value-bind is called on returns only one value....

My guess is that memory is getting corrupted somewhere, but I'm not 
sure.....


FYI the backtrace is
   0: ("FLET SB!IMPL::FAIL" 2)
   1: (NCONC 2)[:EXTERNAL]
   2: ("hairy arg processor for ITERATE::WALK-LIST-NCONCING" (4 I) 
#<FUNCTION ITERATE::WALK> #<FUNCTION "#'(LAMBDA (#:G324 #:G325) (IF # # 
...))" {48A6D9ED}>)
   3: (ITERATE::WALK-ARGLIST (4 I))
   4: (ITERATE::RETURN-CODE-MODIFYING-BODY #<FUNCTION 
ITERATE::WALK-ARGLIST> (4 I) #<FUNCTION "CLOSURE" {48D0732D}>)
   5: (ITERATE::WALK-EXPR (* 4 I))
   6: ("varargs entry for ITERATE::RETURN-COLLECTION-CODE" :VARIABLE NIL 
:EXPRESSION (* 4 I) :START-OPERATION CONS :END-OPERATION (SUBST 
(ITERATE::VAR ITERATE::EXPR) (LIST ITERATE::EXPR)) :ONE-ELEMENT T :TEST 
NIL :PLACE ITERATE::END :RESULT-TYPE LIST)
   7: (ITERATE::APPLY-CLAUSE-FUNCTION ITERATE::CLAUSE-COLLECT-36 
(:COLLECT (* 4 I)))
   8: (ITERATE::PROCESS-CLAUSE (COLLECT (* 4 I)))
   9: ("hairy arg processor for ITERATE::WALK-LIST-NCONCING" ((FOR I 
FROM 0 BELOW 10) (COLLECT (* 4 I))) #<FUNCTION ITERATE::WALK> 
#<FUNCTION "#'(LAMBDA (#:G336 #:G337) (DECLARE #) ...)" {48A6EAB5}>)
   10: ("#'(LAMBDA (#:WHOLE245 #:ENVIRONMENT246) (LET* # #))" (ITER (FOR 
I FROM 0 BELOW 10) (COLLECT (* 4 I))) #S(SB-KERNEL:LEXENV :FUNS NIL 
:VARS NIL :BLOCKS NIL :TAGS NIL :TYPE-RESTRICTIONS NIL :LAMBDA NIL 
:CLEANUP NIL :POLICY ((SPEED . 1) (SPACE . 1) (SAFETY . 1) 
(INHIBIT-WARNINGS . 1) (DEBUG . 1) (COMPILATION-SPEED . 1))))
   ....


--Robbie

>
> Thanks for reporting this,
> -- 
> Andreas Fuchs, <as...@bo...>, as...@ja..., antifuchs
>
>
>
>
> -------------------------------------------------------
> SF.Net is sponsored by: Speed Start Your Linux Apps Now.
> Build and deploy apps & Web services for Linux with
> a free DVD software kit from IBM. Click Now!
> http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click
> _______________________________________________
> Sbcl-devel mailing list
> Sbc...@li...
> https://lists.sourceforge.net/lists/listinfo/sbcl-devel
>