Re: [clisp-list] An internal function segfaults (and "security")

[<don...@is...>]

Sam Steingold writes:
 > oops.
 > looks like we must remove this function too:
 > LISPFUNN(crash,0)
This falls in the category of FFI and OS interface.
It's explicitly meant to get you out of the lisp virtual machine.
Of course, there's also the exit function.
I really think that even these should not directly give you segfaults,
although of course they let you do other things that will.

In any case, I think you (both Dave and Sam) are missing the points
that (1) I'm not asking for every function to check its arguments,
just those on the boundary between lisp and c, and (2) I don't think
that this is expensive.  
BTW, isn't there a way to catch signals at the c level?
Couldn't every otherwise uncaught signal be caught and presented as
a lisp error?  I wouldn't be happy to get a lisp error telling me
about a segfault, but I'd prefer that to exiting lisp.