On Mon, Mar 2, 2015 at 2:02 PM, Matt Oliveri <[email protected]> wrote:
> On Mon, Mar 2, 2015 at 10:26 AM, Jonathan S. Shapiro <[email protected]> > wrote: > > > BitC *never* gets to rely on the optimizer to put things right. > > But Keean never actually said "optimization", Shap. You did. You are right. Keean never uses the term "optimizer". What he does instead is assuming a stack-based rewriting that cannot be accounted for in the semantics of any intermediate form that we commonly understand. He also emphasizes that this rewriting is non-semantic. This makes it sound to me like this is a "compiler internal" rewriting that cannot be expressed within the surface language. To me, all such rewritings are things done by the optimizer. But I also argue against non-compositional typing rules. In general, I agree. There are some corner cases where you can use them without getting into horrible trouble (e.g. disambiguating different uses of '.'), but in general they are very problematic. shap
_______________________________________________ bitc-dev mailing list [email protected] http://www.coyotos.org/mailman/listinfo/bitc-dev
