[jmm-dev] Non-core-model issues status

[Brian Goetz]

> 1. We will spec (outermost) constructors to perform release on exit.
> The spec will not otherwise include specs on final fields, although
> there should be JLS wording to explain consequences; i.e., as one case
> of reads of fields via a reference read for the first time by a thread
> being dependent on that read.  We don't expect to explicitly support
> any further analogs of C++/C11 consume mode (which this may be seen as
> a special case of).

Just a reminder that, with value types coming down the road, we should have a story for if and when release is required on value “constructors” (still searching for the best term for these.)  Gut says: for values that have been declared non-tearable (“volatile”) yes, otherwise no?  

> 2. We introduce and spec a reachabilityFence(Object x) method, and add
> wording about it in JLS sec 12.6. Further suggestions for doing more
> are still welcome. (Also, the method name is still up for grabs.
> Except that the name "keepAlive" is known to be too confusing
> to consider.)

Pardon the bike shed… if the goal here is to communicate liveness info to the GC, then perhaps considering a new form of Reference is the right vehicle, since that’s currently the primary means by which applications can pass notes to the GC after gym class...