This opens an old wound; it was amply discussed in January and I was
originally on the pro-include side of the debate with a large body of
code using it already. I agree that it's convenient; but I've been
convinced it's a bad idea for standardization because, we can frankly
live without it (We avoid the var F = r("f").F pattern almost entirely
in Narwhal by almost always using late binding on var f = r("f")), and
eventually we'll be able to use destructuring like var {F} = r("f")
which is about as convenient *and* safe. It's also become quite
obvious to me that while going from require to include would be easy,
it is *extraordinarily* difficult to do the reverse since information
about which free variables are in use and where they are from is
effectively lost. If that's difficult for the author of a body of
code, it's impossible for someone trying to learn how your code works.
Kris Kowal