[cpp-threads] Preparing for Oxford

[Lawrence Crowl]

On 12/26/06, Herb Sutter <hsutter at microsoft.com> wrote:
> Thinking over the past month's discussion, what effect will/should this have on the C++ MM and atomics proposals for the April meet in Oxford? Here's my suggestion:
>
>   1. Eliminate, or at least split out, the low-level non-SC atomics features.

In the current draft, these features are fairly separable.

>   2. Add type abstractions for a few key uses of non-SC atomics (e.g., DCL, RC).

The current draft has some examples implementing these, but makes
no attempt to formally standardize them.

>   3. (Hard to do for Oxford) Simplify the proposal to use a single ordering relation.

I think at minimum, you need two relations, both control and data
dependences.  (The section also needs to handle sequential code.)  I
was looking at this issue separately, and it appears that you need an
additional weak relation to force the store for an assignment before
a call to a function when the assignment is in the argument list.

-- 
Lawrence Crowl