Re: [boost] Is there any interest in a library for actor programming?

Subject: Re: [boost] Is there any interest in a library for actor programming?
From: Dominik Charousset (dominik.charousset_at_[hidden])
Date: 2013-06-06 05:36:45

On 18.05.2013, at 23:54, Vicente J. Botet Escriba <vicente.botet_at_[hidden]> wrote:
> This is really a lot of good and interesting work on your library.
> 
> Moving all the library to the Boost standards would be quite a lot of 
> work however. Replacing some classes by the corresponding Boost ones.
> 
> I would see already several Boost libraries in your libcppa library:
> * Atoms
> * CowTuples
> * Dynamically Typed Tuples - type erased tuples
> * Tuples Pattern matching
> * Actors

I gave this some thoughts. cow_tuple and any_tuple use the same data representation underneath. In this way, there is no 'conversion' from one to the other. Assigning a cow_tuple to an any_tuple is trivial. To restore the type information, tuple_cast can be used. This cast also allows for wildcards:

auto t0 = make_cow_tuple(1, 2, 3, 4);
any_tuple t1 = t0;
auto opt = tuple_cast<anything, int, int>(t1);
if (opt) {
  auto t2 = *opt;
  cout << get<0>(t2) // will print 3
       << get<1>(t2) // will print 4
       << endl;
}

any_tuple provides the same copy-on-write behavior:

auto t0 = make_cow_tuple(1, "hello", "world"); // const char* is implicitly converted to std::string
any_tuple t1 = t0;
any_tuple t2 = t0;
assert(&t1.get_as<int>(0) == &t2.get_as<int>(0)); // same address
assert(&get<0>(t0) == &t1.get_as<int>(0));        // same address
t1.get_as_mutable<std::string>(2) = "libcppa";
assert(&t1.get_as<int>(0) != &t2.get_as<int>(0)); // no longer the same address
assert(&get<0>(t0) != &t1.get_as<int>(0));        // no longer the same address

Long story short: cow_tuple and any_tuple are closely related. any_tuple *is* a typed erased cow_tuple. Would it be beneficial to split them into two libraries? Any thoughts on that?

Currently, I'm reviewing the tuple classes and evaluate whether boost.serialization/boost::extended_type_info fit the needs of libcppa.

On 05.06.2013, at 15:48, Hartmut Kaiser <hartmut.kaiser_at_[hidden]> wrote:
> Just to reiterate, HPX (https://github.com/STEllAR-GROUP/hpx/) is a very
> similar library exposing actor oriented paradigms to the programmer (amongst
> other things), just wrapped up into a C++11 compliant interface (future,
> async, thread, bind, function, tuple, etc.). It's usable with C++03
> compliant compilers (I have not tried msvc-9.0, though). While not part of
> Boost, it strongly adheres to Boost coding guidelines and style and it is
> maintained by long standing Boost authors.

Hartmut, I can understand that you want to advertise your work, but hpx has nothing to do with actor programming. I saw your presentation at the C++Now conference. hpx offers a future-based API and you are proposing C#'s syntax for async/await (cf. N3564 [1]).

[1] http://www.open-std.org/JTC1/SC22/WG21/docs/papers/2013/n3564.pdf

Best regards,
Dominik