longjmp question

[Jonathan Schilling]

Does the ABI (either C or C++) really need to specify whether longjmp()
destructs auto C++ objects or not?  In the IBM implementations that do
this, is there any extra information passed from compiled code to
setjmp() or longjmp() to support this?  Is there any extra information 
stored in EH tables or data structures to support this?  Are there any
extra interfaces into the EH runtime used by longjmp() to support this?

If not, then the behavior can be implementation-dependent.  Code compiled
by any vendor's compiler either would or would not destruct auto
objects, depending upon which vendor's longjmp() is being used, and/or
which vendor's OS is being used.

Jonathan Schilling		SCO, Inc.		jls at sco.com