[cpp-threads] D2335 (sequential consistency proof) revision

[Boehm, Hans]

I've attached a revision of D2335 (now renumbered D2392, N2335 will not
exist, N2392 is intended tobe in the next mailing, due this weekend).  I
believe this handles try_lock, provided we allow spurious failures.  I
still believe this is the right way to do things, and I haven't heard
any strong disagreement.

I think this makes for a much better story for non-wizard programmers.
Programs that use only locks and SC atomics either are SC, or have a
race under the SC interpretation.  And it doesn't matter whether you
define races in terms of not-happens-before-ordered or simultaneous
execution of conflicting operations by different threads.  The funny
try_lock disclaimer is gone.

Comments are of course most welcome.

Hans
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.decadentplace.org.uk/pipermail/cpp-threads/attachments/20070906/7e3cab15/attachment.html