Andy, below is a small fix which reverts to "standard" xemacs assertion.
I am not sure if it is correct - it turns *off* error checking if
ERROR_CHECK_MALLOC is not defined. Maybe just leave asserts always on,
so that they work in any debug build? Calling abort() without including
lisp.h is not correct because there is special xemacsoid version of abort().
I am constantly getting an assertion in line 673, near the end of malloc().
Do you understand this code to tell what does it mean? xemacs basically
does not work for me, I get it every so often.
Big K
--- gmalloc.c 2000/04/06 21:57:50 1.12.2.4
+++ gmalloc.c 2000/04/07 02:58:21
@@ -4,8 +4,10 @@
/* Get the configuration files if we're being compiled for Emacs. */
#ifdef emacs
# include <config.h>
-#ifdef ERROR_CHECK_MALLOC
-# define assert(x) ((x) ? (void) 0 : (void) abort ())
+# include <lisp.h>
+#ifndef ERROR_CHECK_MALLOC
+# undef assert(x)
+# define assert(x)
#endif
# include "getpagesize.h"
# ifndef HAVE_CONFIG_H
Some time ago, Andy Piper wrote...
This fixes gui-item parsing which was doing the wrong thing for
widgets but
it seems to tickle an extant bug which I am having trouble tracking down in
my cygwin environment. This, for me, causes Edit->Search to crash XEmacs.
If anyone has purify and uses athena and can do a quick run I would be
sincerely grateful.
andy
2000-04-06 Andy Piper <andy(a)xemacs.org>
* gmalloc.c (malloc): add error checking.
(malloc): ditto.
(free): ditto.
(realloc): ditto.
[snip]