On 6/30/16 10:59 AM, jmh530 wrote:
On Thursday, 30 June 2016 at 14:17:22 UTC, Steven Schveighoffer wrote:
Jack Stouffer has created a PR to formalize this.
Please comment if you have objections (the group that argued for
current generate behavior was absent from this post that was meant to
be a debate). I think this is the right thing to do.
It looks like this is just documentation changes. This really doesn't
prevent anyone from making a Range that violates front stability. For
instance, suppose isInputRange is changed so that
auto h = r.front; // can get the front of the range
becomes
auto h = r.front; // can get the front of the range
auto j = r.front;
assert(h == j); //ensure front stability
This doesn't solve it, and it can't be solved -- halting problem. We
have some expectations that are assumed and some that are mechanically
tested. This PR clarifies the assumptions.
-Steve