Search
19:56:26
stassats
(SB-VM::VARYOBJ-PAGE-SCAN-START 16468672) Unhandled memory fault at #x324D8B0.
21:05:28
phoe
What are the steps of making a class object created via DEFCLASS inaccessible so it may be garbage-collected?
21:05:41
phoe
REMOVE-DIRECT-SUBCLASS on STANDARD-OBJECT and SETF FIND-CLASS NIL are not enough.
21:13:18
phoe
I define a finalizer on it that prints "foo". It doesn't get called after I call the GC a few times.
21:15:29
stassats
what else are you doing with it?
21:17:45
phoe
Nothing. I could prepare a test case for it.
21:19:33
phoe
(defclass foo () ()) (sb-mop:finalize-inheritance (find-class 'foo)) (sb-mop:remove-direct-subclass (find-class 'standard-object) (find-class 'foo)) (sb-ext:finalize (find-class 'foo) (lambda () (print "finalized"))) (sb-ext:gc :full t)
21:23:54
phoe
Oh right, there's also (setf (find-class 'foo) nil)
21:25:11
phoe
https://plaster.tymoon.eu/view/1031#1031 <- posting this in the REPL doesn't cause the PRINT to get called.
21:39:47
phoe
hu.dwim.debug:path-to-root seems broken.
21:40:23
phoe
It breaks on its BUILD-REFERENCE-MAP function... which in turn calls SB-VM:MAP-ALLOCATED-OBJECTS.
22:01:32
phoe
stassats: try with which data? Should I fetch new sbcl from sourceforge? or new hu.dwim.debug from somewhere?
22:11:10
phoe
Oh, these are methods. I wonder if it lets me update them on the fly without rebuilding all of SBCL.
22:11:25
phoe
......yes. The finalizer gets called now.
22:15:47
phoe
As for hu.dwim.debug, path-to-root just got removed. Attila told me to use sb-introspect:map-root instead.
22:17:01
stassats
use sb-ext:gc-and-search-roots
22:17:47
phoe
Okay. What is the wps argument supposed to be? The object in question?
22:18:53
phoe
(defvar *foo* (lambda () 42)) (sb-ext:gc-and-search-roots (sb-ext:make-weak-pointer *foo*))
22:19:25
phoe
https://plaster.tymoon.eu/view/1032#1032 <- my inferior Lisp buffer says this.
22:19:29
phoe
And 0 gets returned from the function.
22:19:47
phoe
Ooh. I need to do it from a non-threaded Lisp?
22:19:59
phoe
I mean, without any other threads running?
22:20:15
stassats
no, it'll work in slime, but will show nothing in the repl
22:21:19
phoe
Still tells me there's a failure tracing.
22:21:57
phoe
https://plaster.tymoon.eu/view/1033#1033
22:22:20
phoe
That is SBCL ran from the terminal.
22:30:28
phoe
Is this expected, or another quirk I found?
23:28:11
stassats
i don't have the slightest idea
23:28:18
stassats
works fine on not-functions
23:29:48
phoe
I'll make a ticket for that
0:46:30
aeth
Is there a way to get sb-mpfr to print normally like 3.1415926535897931e0 instead of starting with the . like .31415926535897931e+1?
0:47:57
aeth
It looks like I'll have to write my own print.
2:45:17
pfdietz
Pushed my working copy of ansi-test to github. Usability not much there right now.
2:45:36
pfdietz
I intended to asdf-ify it soonish.