25 Apr
2005
25 Apr
'05
9:23 a.m.
Peter Dimov wrote: [...]
As for the "afford to release"... There's always the safe choice of <cpthread> + specifying that cancellation is a C++ exception. C++ification can then be done in a TR.
That shall better be done first, before "fast tracking" thin OO layer of questionable value. C++ shall simply incorporate Pthreads and provide both legacy C interface and new better OO/GP stuff. Ultimately, some TR or whatnot shall provide reference implementation of legacy stuff using new <thread>, not the other way around. ADA-like highlevel stuff a la protected objects is the next step (C++1X, I'd guess). regards, alexander. -- http://www.google.de/groups?selm=412533CA.82703A04%40web.de