On Sat, Jan 18, 2014 at 7:18 AM, Jonathan S. Shapiro <[email protected]> wrote:
> But I begin to suspect that, in the absence of downcast, there are other > graph algorithms that are more difficult. Can anybody identify a case to > think about where (a) the ability to downcast really appears to be > essential, and (b) the set of underlying types is not closed? If the set of > underlying types is closed then I know how to handle downcast within > BitC-current. It's nasty, but I know how to do it. If the set is open (think > extensibility here), then RTTI becomes essential and downcast has to be done > as a language primitive (because it needs to guarantee safety of the > underlying runtime primitive). > > Can anybody come up with a good example to think about as a case study here? Graph Rewriting was the the only thing i could really think of. _______________________________________________ bitc-dev mailing list [email protected] http://www.coyotos.org/mailman/listinfo/bitc-dev
