Matt Calabrese <rivorus <at> gmail.com> writes:
On Thu, Mar 5, 2015 at 6:41 AM, Louis Dionne <ldionne.2 <at> gmail.com> wrote:
[...]
I'm sure these problems can be resolved, but I definitely think it is important to fully address them before formal review. The subtlety of the problems and the ease of misuse makes things really scary to me and I would be worried if a lot of library code ended up getting written on top of this without solutions in place.
If any of this is already solved by way of tricks that I'm not aware of (I haven't looked through the source yet), then just make sure people know that they, too, must utilitze such tricks when making their own hana-style metafunctions.
Yes, this is already taken care of and no unfortunate instantiations will take place. The downside is that `decltype(tuple_t<T...>)` is an unspecified dependent type that only _inherits_ `_tuple<T...>`. It is still possible to use pattern matching in overloads as follows: template <typename ...T> void f(hana::_tuple_t<T...>) { // ... } You can see how this is implemented here: http://goo.gl/iISNpb And the unit test here: http://goo.gl/sBqMZo I'm playing with fire, I know :-). I read the process of ADL very carefully when I thought about this, but it is also possible that I'm exploiting a Clang bug. What do you think? Louis P.S.: Thanks for letting me know of this issue at C++Now.