Re: Computation of exception handling

["Martin Dowie" <martin.dowie@gecm.com>]

Thanks for the reply - I'm one of those waiting for the ravenscar support
:-)

Peter Amey <pna@praxis-cs.co.uk> wrote in message
news:3A7E6C0B.5DF54BA@praxis-cs.co.uk...
> It certainly looks ok at first glance.  The difficulty from our point of
> view is that it is usually very easy to come up with examples of safe
> use of almost any language feature.  (Robert is fond of pointing out how
> useful goto can be if used sensibly).  From the SPARK design point of
> view the problem is not showing that something can be safely used but
> showing that there is _no_ way of using that creates ambiguous
> behaviour.  We are in the "for all constructs no ambiguity" rather than
> "there exists a safe use" territory!
>
> I am sure that we could come up with something for local exception
> handling; however, feedback from users suggests that this is much less
> important than the things we are doing to support "abstract oriented
> programming" such as the INFORMED design methods, using annotations to
> bind _system_ and _software_ designs, proof of programs involving
> abstract state and support for Ravenscar.
>
> Peter