On Wednesday 11 February 2015 11:11:36 Olivier Goffart wrote: > "UB could ckick in" has no meaning. > > In practice there is no reason why casting a pointer to member function to > remove the const would not work. Yet, you would not accept it[1]. > > Data races are undefined behavior according to the standard. So anything > could happen.
I believe the N4323 paper Thiago cited is meant as a step in the direction of tightening (loosening?) the rules around the definition of data races, so writing the same value would be legal even if the writes conflict. But the summary doesn't say so, and I didn't have time yet to read the full paper. If and when the execution model is thus changed, we can assume that older compilers will respect it, too, and use it. But I agree we can't until then. -- Marc Mutz <[email protected]> | Senior Software Engineer KDAB (Deutschland) GmbH & Co.KG, a KDAB Group Company www.kdab.com || Germany +49-30-521325470 || Sweden (HQ) +46-563-540090 KDAB - Qt Experts - Platform-Independent Software Solutions _______________________________________________ Development mailing list [email protected] http://lists.qt-project.org/mailman/listinfo/development
