C++ memory model

[Doug Lea]

I think that's all the meta-chat we are allowed on this issue :-).

How about proposing something? Not necessarily in excruciating
detail. But something to consider as an alternative or supplement to
Hans's and my  proposal to have the bottom levels be the full
fleshing-out of Hans's partial fleshing out of the approach in the
original proposal? (OK, that sentence sounds stupid, but we know what it
means!)

Beware that if you want to propose that the only supported concurrent
programs are ones that use locks, the non-blocking sync folks won't like
you.

-Doug