Re: [boost] Re: Boost.Threads - thread_desc concept

From: John Max Skaller (skaller_at_[hidden])
Date: 2001-06-28 16:40:38

Peter Dimov wrote:

> > A thread blocked waiting for input
> > cannot always detect such conditions (only an actual
> > input error). In such cases, cancellation is not
> > a bad idea, it is mandatory.
> 
> But does thread cancellation unwind the thread stack? If it does, how is
> this possible to implement without core language support?

        Exactly! We must have core language support.
I've been saying this all along. Whether or not there
is a Standard 'thread' library for C++ is of secondary
importance. Updating the abstract machine so that
it is possible to resolve these issues in the core
language is primary. The resolution may well be
'implementation defined' or 'undefined behaviour',
but without a resolution to various issues,
all threaded programs are beyond the scope of the
Standard (and of course a Standard thread library is
impossible).

        I support the proposal of a thread library
because it will _force_ proposal of changes to
both the conformance model (abstract machine etc)
and core language semantics, so that in fact,
many existing programs might be closer to
being called C++.

-- 
John (Max) Skaller, mailto:skaller_at_[hidden] 
10/1 Toxteth Rd Glebe NSW 2037 Australia voice: 61-2-9660-0850
New generation programming language Felix  http://felix.sourceforge.net
Literate Programming tool Interscript     
http://Interscript.sourceforge.net