Custom Query (1030 matches)

Using cl-dispatch, I perform the elementary operation of adding a callback function to the asynchronous dispatch queue. See code below. Thread crash results when testing by repeated execution, typically after tens of executions. The error details are slightly different each time, and occasionally I am thrown into the kernel debugger. I have attached the OS X (10.7) crash report for one crash.

(load "lisp:quicklisp;setup")
(ql:quickload "dispatch")

(defcallback foo (:address x) (declare (ignore x)) )

(defun test (v)
  (let ((q (#_dispatch_get_global_queue 0 0))) 
    (dotimes (i v)
        (print i) 
       (#_dispatch_async_f q (%null-ptr) foo))))

(test 100)   ; this never completes without an error.

Using &WHOLE in the lambda list for the :ARGUMENTS section of the long form of DEFINE-METHOD-COMBINATION results in an UNBOUND-VARIABLE condition. Here's a simple test case:

CL-USER> (define-method-combination test ()
             ((primary () :required t))
           (:arguments &whole whole)
           (format t "this is it: ~A" whole)
           `(call-method ,(first primary) ,(rest primary)))
;Compiler warnings :
;   In an anonymous lambda form: Undeclared free variable #:ARGS
TEST
CL-USER> (defgeneric foo (x)
           (:method-combination test)
           (:method (x) x))
#<STANDARD-GENERIC-FUNCTION FOO #x30200127BC6F>
CL-USER> (foo 3)

You can see the warning when the method combination is defined, then the call results in:

Unbound variable: #:ARGS
   [Condition of type UNBOUND-VARIABLE]

This is on CCL Version 1.8-dev-r15225M-trunk (DarwinX8664)

An extra "." can be printed when *print-circle* is t. A log is below. The version (15230) is the trunk as of two days ago. Not sure how to specify the "component" -- I'd guess it's a source code bug, but I don't know.

Welcome to Clozure Common Lisp Version 1.8-dev-r15230M-trunk  (LinuxX8664)!
? (let* ((*print-circle* t)
	 (atm "hi")
	 (alist `((3 . ,atm) (2 . ,atm))))
    (print alist))

((3 . #1= . "hi") (2 . #1#)) 
((3 . "hi") (2 . "hi"))
? '((3 . #1= . "hi") (2 . #1#))  ; the extra dot ruins readability, of course:

> Error: Reader error on #<CCL::RECORDING-CHARACTER-INPUT-STREAM #x3020006EDE3D>, near position 13:
>        Dot context error in ".".
> While executing: CCL::SIGNAL-READER-ERROR, in process listener(1).
> Type :POP to abort, :R for a list of available restarts.
> Type :? for other options.
1 > :q
? '((3 . #1="hi") (2 . #1#)) ; manual correction
((3 . "hi") (2 . "hi"))
?