Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cornell!murthy
From: murthy@rati.cs.cornell.edu (Chet Murthy)
Newsgroups: gnu.emacs.bug
Subject: GNUemacs bug
Message-ID: <26278@cornell.UUCP>
Date: 21 Mar 89 00:33:20 GMT
Sender: nobody@cornell.UUCP
Reply-To: murthy@cs.cornell.edu (Chet Murthy)
Distribution: gnu
Organization: Cornell Univ. CS Dept, Ithaca NY
Lines: 33


Bug Report:

Version: GNUemacs version 18.52 with X11 support

Synopsis:

After booting Emacs, type in, to the *scratch* buffer:
-----begin--------
(setq x (cons nil nil))
(progn (rplaca x x) (rplacd x x) 1)
(setq stack-trace-on-error t)
(insert-string x)
------end---------

At this point, Emacs will dump core.  The bug seems to be that the
backtrace printing routines don't know about infinite objecs, so they
die a grisly death when presented with such.

Another way to get the bug to occur is with the following code:
----begin----
(defun foo (x) (error "x"))
(setq x (cons nil nil))
(setq stack-trace-on-error t)
(progn (rplaca x x) (rplacd x x) (foo x))
---end-------

I really don't know where in the code to go rooting around for such an
error, though I intend to find out.


	--chet--
	murthy@svax.cs.cornell.edu