Opened 12 years ago

Closed 12 years ago

#201 closed defect (fixed)

&WHOLE broken in DEFINE-METHOD-COMBINATION

Reported by: Greg Owned by: somebody
Priority: minor Milestone: 1.1.0
Component: CLOS Version: 1.1.0-dev
Keywords: ansi-conformance Cc:
Parent Tickets:

Description

I get an error whenever I call a function that uses the simple method combination shown here.

CL-USER> (define-method-combination progn-with-whole ()
           ((methods ()))
           (:arguments &whole whole)
           `(progn (format nil "using ~a" ,whole)
                   ,@(mapcar (lambda (method) `(call-method ,method))
                             methods)))
PROGN-WITH-WHOLE
CL-USER> (defgeneric test-&whole (x)
           (:method-combination progn-with-whole)
           (:method (x) x))
#<STANDARD-GENERIC-FUNCTION TEST-&WHOLE {34315D04}>
CL-USER> (test-&whole 4)

Here is the error:

The value 4 is not of type LIST.
   [Condition of type TYPE-ERROR]

Restarts:
 0: [RETRY] Retry SLIME REPL evaluation request.
 1: [*ABORT] Return to SLIME's top level.
 2: [ABORT] Abort thread.

Backtrace:
  0: (#<FUNCTION {C69203}> #<TYPE-ERROR {FE85ADF}> #<FUNCTION {C69203}>)
  1: (APPLY #<FUNCTION {C69203}> (#<TYPE-ERROR {FE85ADF}> #<FUNCTION {C69203}>))
  2: (SYSTEM::RUN-HOOK SYSTEM::*INVOKE-DEBUGGER-HOOK* #<TYPE-ERROR {FE85ADF}> #<FUNCTION {C69203}>)
  3: (INVOKE-DEBUGGER #<TYPE-ERROR {FE85ADF}>)
  4: org.armedbear.lisp.Lisp.error(Lisp.java:379)
  5: org.armedbear.lisp.Lisp.type_error(Lisp.java:420)
  6: org.armedbear.lisp.LispObject.car(LispObject.java:156)
  7: org.armedbear.lisp.Lisp.eval(Lisp.java:520)
  8: org.armedbear.lisp.Lisp.evalCall(Lisp.java:559)
  9: org.armedbear.lisp.Lisp.eval(Lisp.java:515)
 10: org.armedbear.lisp.Lisp.progn(Lisp.java:684)
 11: org.armedbear.lisp.SpecialOperators$sf_progn.execute(SpecialOperators.java:275)
 12: org.armedbear.lisp.Lisp.eval(Lisp.java:505)
 13: org.armedbear.lisp.Lisp.progn(Lisp.java:684)
 14: org.armedbear.lisp.Primitives$sf_macrolet.execute(Primitives.java:3641)
 15: ((LAMBDA (#:G13665) (MACROLET (#) (PROGN # #))) (4))
 16: (FUNCALL (LAMBDA (#:G13665) (MACROLET (#) (PROGN # #))) (4))
 17: (#<STANDARD-GENERIC-FUNCTION TEST-&WHOLE {34315D04}> 4)
 18: (APPLY #<STANDARD-GENERIC-FUNCTION TEST-&WHOLE {34315D04}> (4))
 19: (MOP::INITIAL-DISCRIMINATING-FUNCTION #<STANDARD-GENERIC-FUNCTION TEST-&WHOLE {34315D04}> (4))
 20: (#<STANDARD-GENERIC-FUNCTION TEST-&WHOLE {34315D04}> 4)

Change History (5)

comment:1 Changed 12 years ago by Mark Evenson

Milestone: 1.1.0

comment:2 Changed 12 years ago by Mark Evenson

Keywords: ansi-conformance added

comment:3 Changed 12 years ago by ehuelsmann

Resolution: invalid
Status: newclosed

Well, of course the error can be improved, but the code provided is in error. The backquote form should have read:

`(progn (format nil "using ~a" ',whole))

Note the quote before the unquote. The difference being that the new code evaluates to

(progn (format nil "using ~a" '(4))

which takes (4) as a constant while the old code does to

(progn (format nil "using ~a" (4))

Which tries to evaluate a function by the name 4.

Resolving as invalid. Although I appreciate the fact that the error could be better, I don't see how at this time. (The problem being that LispObject.car() raises a type error "This is not a LIST", while the outer code knows which error *should* have been generated, but has no way to communicate that to inner code without presumed-large performance impact.

comment:4 Changed 12 years ago by ehuelsmann

Resolution: invalid
Status: closedreopened

Reopening. On re-reading, the WHOLE variable should have been bound to a form evaluating to (4), so the ticket is valid and I need to figure out how to do this.

comment:5 Changed 12 years ago by ehuelsmann

Resolution: fixed
Status: reopenedclosed

The commits r14040, r14041, r14043, r14045, r14046, r14052, r14053, , r14054, , r14055, r14058 and r14059 together should fix this problem and several others which existed due to my misunderstanding of the spec.

Closing this issue as fixed now. Please test the implementation thoroughly.

Note: See TracTickets for help on using tickets.