Andrew Sutton

Fixed in the injection-2 branch.

This hasn't been an issue because we've only been using `idexpr` within metaprograms. Those are not dependent when instantiated -- or at least they haven't been yet.

This is related to #25. Fixing that will fix this also.

This is partially implemented via the `typename` operator. However, it needs to be more fully tested, especially as it appears in templates and it should probably be supported in nested...

Known issue. Requires a lot of work to fix it. On Mon, Jan 16, 2017 at 1:44 PM Jennifer Yao wrote: > Despite __reflect_specifiers() being a recognized reflection trait >...

I think this might be computing the symbol address and then dereferencing that. We probably want to extract the variable or string computed by the immediate call and then return...

Maybe: "An object is well-formed if it does not have indeterminate value". There are some exceptions about the validity of the use of indeterminate value for narrow character types: http://en.cppreference.com/w/cpp/language/default_initialization...

> Is a NaN an indeterminate value? In what sense? It isn't, and trying to exclude it will get you into a long discussion with Lawrence Crowl. There are (probably)...

> So, double does not satisfy EqualityComparable or TotallyOrdered? Can you pass NaNs to sort? No. Can you pass them to an algorithm designed to deal with them while comparing...

> It is a bit more complex than that - I can have two well-formed iterators > that are not equality comparable but that doesn't mean that iterators are >...