Joel de Guzman writes:
> David Abrahams wrote:
>> Joel de Guzman writes:
>>> BTW, thanks for the patch. I'll apply later today. There's some
>>> overlap with actor_result.hpp. I'll try to merge or unify both
>>> solutions. Doing so should simplify the code. I'm still trying
>>> to understand the need for typename add_const::type. Could
>>> you explain a bit?
>> It handles rvalues correctly. When you pass a non-const rvalue to an
>> actor it will be treated as a const&... Oh, wait, you're not handling
>> the forwarding problem, are you? In that case, maybe you can't pass
>> non-const rvalues to actors?
> No, it's the same as lambda. Function forwarding problem is not
> handled. Is there a new trick I don't know yet that does not have
> an N^2 combinatorial explosion?
No, you just generate the combinatorial overloads. It isn't that bad
for reasonable N, really. See enclosed.