[PATCH 21.5] assert_failed should not return

[Stephen J. Turnbull]

Jerry James writes:

 > PATCH 21.5

 > FWIW, emacs behaves in the way I consider sensible.  If you trigger an
 > assertion failure while already handling an assertion failure or fatal
 > error, emacs shuts down immediately.

Uh, in practice, so does XEmacs.  Only if somebody insane (or who is
Ben Wing, which may not be the same thing ;-) #defines
ASSERTIONS_DONT_ABORT does this extremely dangerous behavior occur.
But that requires actually editing the source AFAIK.  Are you sure
you've actually observed assert_failed returning?

XEmacs *did* (or does) have a different issue, where signals during
the shutdown process are handled rather than dumping core.  If you've
actually observed recursive crashes, I don't think the change you're
making will prevent them.