longjmp question

[John Wilkinson]

Jim Dehnert wrote:
> 
> I'll add the issue.  But...
> 
> Implementation is a significant issue.  The normal longjmp implementation
> is very simple -- setjmp stores the register/stack state, and longjmp
> copies it back and branches.  There is normally no traceback involved,
> so what you suggest is a dramatic change, and probably would make C
> people very unhappy.  Furthermore, C++ users have the option of using
> C++ exceptions, which have the effect you seek.
> 
> -           Jim Dehnert         dehnert at sgi.com
>                                 (650)933-4272

The C++ standard, 18.7 paragraph 4, says a call to longjmp has undefined
behavior if any automatic objects would have been destroyed by a
throw/catch with the same source and destination.   I don't see that
this is something we need to fix.


-- 
John Wilkinson