A collection of DIPs

[via Digitalmars-d]

On Monday, 7 September 2015 at 14:52:21 UTC, Ola Fosheim Grøstad 
wrote:
> 1. You need to make sure that pointers to the interior of a GC 
> object are not live when the last pointer to the start of the 
> GC objects disappears.
>
> 2. You need to annotate C functions that take pointers with a 
> guarantee that they don't hold on to references through that 
> pointer. Or simply ban C functions from taking GC memory.

3. You need give up on destructors for GC objects too, and not 
allow them to own heap allocated objects. A reasonable 
restriction IMHO.