Hi all,
I'm trying to port some software I had written that works fine on
allegro CL to cmucl. I'm getting an unusual error: a form that runs fine
at the top level fails to run when I have it in a function, and execute
the function. Before I try to produce a bug report, I'd like to try
things on a relatively new CMUCL snapshot. I'm presently working with
the april snapshot that used to be on common-lisp.net and has since
vanished. Is there a newer snapshot available?
Also, I'll need a bit of advice on how I might track down this bug.
Here's the function that appears to be breaking:
(defun define-stack-test-suite ()
(defsuite stack-object-tests ()
((stack :initform nil))
;; Let's test a stack
(:setup (setq stack nil))
(:test () (null stack))))
defsuite, as you might guess, is a macro that expands ultimately into a
few definitions, including classes and methods. When I run this
function, I see this:
FRET> (define-stack-test-suite)
; Compiling LAMBDA (.PV-CELL. .NEXT-METHOD-CALL. SI):
; Compiling Top-Level Form:
; Compiling LAMBDA (.PV-CELL. .NEXT-METHOD-CALL. OBJ):
; Compiling Top-Level Form:
; Compiling LAMBDA (.PV-CELL. .NEXT-METHOD-CALL. SI):
; Compiling Top-Level Form:
Execution of a form compiled with errors:
(DEFCASE (STACK-OBJECT-TESTS) (NULL STACK))
Error flushed ...
; Evaluation aborted
Not particularly useful. Also an interesting outcome of running my code
is...
CL-USER> (quit)
Error in function UNIX::SIGSEGV-HANDLER: Segmentation Violation at
#x100000C8.
Error flushed ...
; Evaluation aborted
BTW, I also have slime loaded. Without slime, the behavior is a bit
different. I can actually see more of the error condition.
; In: FRET:DEFSUITE FRET::STACK-OBJECT-TESTS
; (FRET:DEFSUITE FRET::STACK-OBJECT-TESTS NIL (#) (:SETUP #) ...)
; --> PROGN
; ==>
; (FRET:DEFCASE (FRET::STACK-OBJECT-TESTS) (NULL FRET::STACK))
; Error: (during macroexpansion)
; Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER:
; NIL is not of type FRET::TEST-SUITE-DATA
; [GC threshold exceeded with 18,936,864 bytes in use. Commencing GC.]
; [GC completed with 11,213,568 bytes retained and 7,723,296 bytes freed.]
; [GC will next occur when at least 23,213,568 bytes are in use.]
Test Case Failure: #<FRET::SUITE-IO-TESTS-TEST-CASE-2 {5884D37D}>
Error in function UNIX::SIGSEGV-HANDLER: Segmentation Violation at
#x99E5A513.
[Condition of type SIMPLE-ERROR]
Restarts:
0: [RETRY ] Retry this test.
1: [SKIP ] Skip this test.
2: [DISABLE] Disable this test and continue.
3: [ABORT ] Return to Top-Level.
Debug (type H for help)
(UNIX::SIGSEGV-HANDLER #<unused-arg>
#<unused-arg>
#.(SYSTEM:INT-SAP #x3FFFECA8))
Source: Error finding source:
Error in function DEBUG::GET-FILE-TOP-LEVEL-FORM: Source file no longer
exists:
target:code/signal.lisp.
0] backtrace
0: (UNIX::SIGSEGV-HANDLER #<unused-arg>
#<unused-arg>
#.(SYSTEM:INT-SAP #x3FFFECA8))
1: (UNIX::SIGSEGV-HANDLER 3
#<unused-arg>
#<unused-arg>
#.(SYSTEM:INT-SAP #x3FFFECA8))[:EXTERNAL]
Again, doesn't tell me all that much, except that I might be tickling a
bug in CMUCL internals. I'm running on x86, fedora fc2, on a dual
processor box though I doubt this is relevant. Any suggestions on
debugging this problem?
If anyone wishes to try out the source code, you can either fetch it
using svn from http://www.mono-picante.org/vc/fret/trunk/, or if you
don't I can post it.
Thanks!
Sunil
