Sharing in D

[Walter Bright]

Sean Kelly wrote:
>> Sure, it won't be pretty, but there won't be any ambiguity of the order
>> in which things happen.
> 
> A totally worthwhile tradeoff IMO.  The alternative would be to
> have ++ call INC if supported or a CAS loop if not, and that's
> a hidden cost that may surprise users.  It could always be
> enabled later if needed anyway.

I'm surprised, we expected much more flak about that <g>. The idea is to 
screw everything down tight like this, see how it goes, and then 
experience will tell us where we need to loosen things up.

> Oh, and darn you for getting me moderately enthusiastic for
> D 2.0 :-)

I for one am very enthusiastic about the new threading model. The real 
validation would be when other languages start copying it!