Re: [boost] Re: Ending/canceling/suspending a thread (proposal)

From: Peter Dimov (pdimov_at_[hidden])
Date: 2001-08-23 10:57:39

From: <williamkempf_at_[hidden]>
> --- In boost_at_y..., "Peter Dimov" <pdimov_at_m...> wrote:
> > Ending a thread:
> >
> > No library support is necessary.
>
> Uhmm... you describe the necessary library support below ;).

:-)

I meant that no special thread::exit() function is necessary.

> Just curious, what would a derived type be good for?  I'd expect that
> thread::cancel (and maybe thread::exit) would be the only methods
> that could throw this exception so we wouldn't have derived types any
> way.

Well, a catch() handler for thread_end will always catch derived types as
well; right now I don't see any use for a derived type except not
over-constraining implementations. When canceling a thread, the
implementation should probably be allowed to throw different exceptions in
different situations, as long as they are derived from thread_end.

> You *MUST* include the ability to change cancelability for a thread.

Yes, you're probably right. This is a 'phase 0' proposal.

> > Suspending/resuming a thread:
>
> I'm not sure we should include these.

I'm not sure either. The methods are safer when a suspended thread is
treated as a resource, and the suspend/resume calls are wrapped in a RAII
object, similar to scoped_lock. That's why thread::resume is not a
cancelation point. :-)

Still, deadlocks are possible.

> > thread::suspend is a cancelation point.
>
> Why?  It's not a blocking operation for the thread calling it, so
> making this a cancelation point doesn't make much sense.

thread::suspend() accesses a shared object - the thread state - and can,
therefore, be a blocking operation since the access will need to be
serialized. At least that's how I look at it; I may be wrong.

> > Opinions? This behavior (with one minor exception) is implementable
> now, at
> > least on paper.
>
> What's the minor exception?

Well you have identified it yourself, it's catching thread_end escaping from
main(). For now we'll probably have to provide thread_main() instead. :-)

--
Peter Dimov
Multi Media Ltd.