From: Martin Wille (mw8329_at_[hidden])
Date: 2005-08-17 10:53:56
Thorsten Schuett wrote:
> On Wednesday 17 August 2005 16:06, Martin Wille wrote:
>>I think that would be viable. Another approach would be to pass a
>>combiner (similar to what Boost.Signals does) and to use
>>optional<whatever> as return types.
optional<> would be insufficient if we don't have only success/failure
but some quality indicator. We could return a result if one of the
futures returns a result with quality == "best". Otherwise, we have to
wait until all the futures involved have returned a result.
(Something like that would be useful for running several heuristic
algorithms in parallel with a suitable definition of "best")
Send instant messages to your online friends http://au.messenger.yahoo.com
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk