At Tuesday 2004-02-10 03:21, you wrote:
"Victor A. Wagner, Jr." wrote: [...]
I'm convinced that propagating *unexpected* exceptions into the joining thread is totally wrong and insensible.
would you rather abort the application??
Absolutely.
fine, don't catch any exceptions when you do the join(), your system will crash gloriously.
That said, feel free to catch anything/everything but please don't impose this silliness on me. The only problem with respect to threads/futures is that the dynamic context in which exceptions are thrown (in the "joinee" thread) is "detached" (in space and time) from the dynamic context that is supposed to "decide" (my means of try handler(s) or lack thereof) whether this or that exception is expected or unexpected.
I suggest again, that you view the thread creation and later join as the dispatch / results portion of a normal function call. If you cannot see that as at least one possible use of threads, then there isn't any point in further discussion with you.
That leads to the conclusion that a typelist of expected (to be caught in the joinee thread and "re-rasied" in the joiner thread at join point) exceptions shall be specified at thread/future creation time.
I submit the argument against throw specifications (see all the relevant literature).
regards, alexander.
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Victor A. Wagner Jr. http://rudbek.com The five most dangerous words in the English language: "There oughta be a law"