OK. I've gone and grabbed bitc-lang.blogspot.com (bitc was taken), but I don't have time to set it up properly this morning.
On Tue, Jul 8, 2014 at 7:38 AM, Jonathan S. Shapiro <[email protected]> wrote: > Raoul: > > I'm afraid you shouldn't hold your breath on this request, but it may help > that the things you are asking about are mostly orthogonal. > > The slippery slopes in subtyping mainly have to do with interactions > between subtyping and type inference. There is a *huge* research > literature on this, and the devil is in the details. There are several ways > to dodge or mitigate the bullets. We're not far enough along in BitC to > know whether those will work for us yet, which is part of my hesitation. I > also hesitate because once you commit yourself to dodging a bullet in a > particular way, that choice will tend to constrain the future ways in which > the type system of the language can evolve. > > If you want to try to get a feel for this, the best source might be > Benjamin Pierce's book Types and Programming Languages. If you're more > inclined toward academic papers: > > Simple, Decidable Type Inference with Subtyping > <https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=3&cad=rja&uact=8&ved=0CDMQFjAC&url=http%3A%2F%2Farxiv-web3.library.cornell.edu%2Fpdf%2F1104.3116v1&ei=fgC8U97rGar-igLnIQ&usg=AFQjCNHBxWy_UoSWCzi0aUNwglJY7zIr6w&sig2=rOzL6Uks2pKCTwU1ARpqeQ&bvm=bv.70138588,d.cGE> > > > My personal expectation is that the real killer for inference in BitC is > going to turn out to be something called kinding. That's an issue that's > considered too advanced for TAPL to address. > > > As to interfaces, I'm really only aware of two kinds (ignoring BitC): in > some legacy languages an interface is really a module boundary. Otherwise, > the Java/C# notion of interfaces seems to be how people do things. > > BitC interfaces are probably misnamed. What they really do is existential > encapsulation, which is why they were originally named "capsules". At some > point I noticed that they looked a whole lot like interfaces (except for > the downcast and identity issues) and started calling them that. The > discussion here on the list strongly suggests that calling them > "interfaces" probably isn't a good idea. > > > And yes, BitC needs a blog (or better: a CMS). I haven't put one up on my > machines because of security concerns, and because I don't want the blog to > detract from the discussions here. It may be time to start one, though. > > > Not sure how helpful this is... > > shap > >
_______________________________________________ bitc-dev mailing list [email protected] http://www.coyotos.org/mailman/listinfo/bitc-dev
