Segfault in change-class (was Re: [Gcl-devel] defgeneric method fix)

[Paul F. Dietz]

Camm Maguire wrote:

> Hi Paul!  The patch below will get you past the 'method not working
> yet' issue.  Now the tests stop with the messages below.  Insights
> appreciated. 

I don't know what's causing this error (it's deep in tbe bowels of PCL,
using the second incarnation of find-method-combination).  I've worked
around it by surrounding the various top level error forms with
a macro REPORT-AND-IGNORE-ERRORS.  This will cause tests that depend
on these generic functions/methods/etc. to fail, but at least it lets
us make progress elsewhere.

I am now getting a segfault when running the tests.  I've tracked
it down to a test in change-class.lsp, specifically change-class.error.5.
This test checks that an error is thrown if an attempt is made
to change an object into an instance of a built in class.  Here's
an example showing the bug:

GCL (GNU Common Lisp)  (2.5.3) Fri Jul 11 21:49:10 CDT 2003
Licensed under GNU Library General Public License
Dedicated to the memory of W. Schelter

Use (help) to get some basic information on how to use GCL.

>(defclass foo () ())

#<Standard-Class FOO 1034505734>

>(loop for i from 0 do (print i) do (ignore-errors (change-class (make-instance 'foo) (class-of 
#\Space))))

0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Unrecoverable error: Segmentation violation..

(I guess it is corrupting something and died later.)

        Paul