On Nov 25, 2013, at 3:25 PM, Carl Eastlund wrote: > I agree. It doesn't bind in the sense of extending an environment, but it > does in the sense of causing a new set of references to be resolved, e.g. > x.method_name() for any x that now implements the trait in question.
Hmm... It's not really worth arguing about, but I think that this is a stretch. It seems clear to me (I guess it would, wouldn't it?) that the binding-ness or non-binding-ness is distinctly secondary here to the simple fact that I want to abstract over something that's not in an expression position, and thus can't be replaced by a function call. I suppose I should come up with more examples. John > > Carl Eastlund > > > On Mon, Nov 25, 2013 at 4:35 PM, Matthias Felleisen <matth...@ccs.neu.edu> > wrote: > > On Nov 25, 2013, at 12:54 PM, John Clements <cleme...@brinckerhoff.org> wrote: > > > they're not binding any new identifiers; they're just declaring that this > > type implements this trait. > > > Thanks for the clarification. I still think this category and binding should > be merged into 'says something about an identifier' -- Matthias > > > ____________________ > Racket Users list: > http://lists.racket-lang.org/users > > ____________________ Racket Users list: http://lists.racket-lang.org/users
