From: bill_kempf (williamkempf_at_[hidden])
Date: 2002-01-17 12:56:40
--- In boost_at_y..., "David Abrahams" <david.abrahams_at_r...> wrote:
>
> ----- Original Message -----
> From: "terekhov" <terekhov_at_d...>
>
>
> > --- In boost_at_y..., "Peter Dimov" <pdimov_at_m...> wrote:
> > [...]
> > > During stack unwinding, stopping the cancel exception is
mandatory,
> > > otherwise the whole process is going to die. For example a C++
> > exception may
> > > be on its way up when someone cancels the thread.
> >
> > But that is N6, unless I am missing something:
> >
> > ">6) Ensure cancellation points don't throw when
uncaught_exception()
> > >returns true."
>
> Note: uncaught_exception returns false in catch blocks, but catch
blocks
> normally contain the same sort of code as destructors (this is why
Herb
> doesn't like it).
I'm not sure I know what you are trying to illustrate here. I agree
with the above statement, but in our case you can at least deal with
cancellation in destructors (by turning it off or using try/catch),
while during stack unwinding we could cause the application to
terminate.
Are you against N6?
Bill Kempf