Re: [asio] using fork in coroutine composed operations

From: Stephan Menzel (stephan.menzel_at_[hidden])
Date: 2019-09-09 10:59:35

Am Mo., 9. Sept. 2019 um 09:03 Uhr schrieb Gavin Lambert via Boost-users <
boost-users_at_[hidden]>:

>
> > It can be called without an argument (for kick-off), with a single
> error_code (from async_wait)  or with two argiments (from async_read).
>
> Yes, that's the method the OP is currently using.  It requires using a
> coroutine or other state to determine which "step" you're up to, whereas
> with separate overloads this is inherent in which method is being called.
>
> Neither approach is necessarily "better" than the other; it depends what
> you're doing in there as to which one ends up being more readable.
>

Well, I got both implementations (async timed connect and socks4 handshake
in place now and I do prefer the coroutine approach as it is much easier to
read and I wanted to adapt this to more functions very soon.

The problem really are the different handler signatures. Those default
arguments are not a silver bullet by any means. I have managed to get it
done using beast::bind_handler() in my timed_connect impl but pretty much
the identical code failed entirely in the socks4 handshake.
I can now only assume that using bind_handler() with the move_ahead pattern
simply doesn't work and without any kind of bind_handler you will quickly
run out of ways to order the arguments in operator() to fit all functions.

In any case, I am glad I got the impl as far as this and that's a great
step forward in re-using code in my projects.

Thanks for all your suggestions,

Stephan