Re: [boost] needs coroutine2 a review?

Subject: Re: [boost] needs coroutine2 a review?
From: Andrey Semashev (andrey.semashev_at_[hidden])
Date: 2015-06-21 05:31:30

On 21.06.2015 12:05, Vicente J. Botet Escriba wrote:
> Le 02/06/15 11:00, Oliver Kowalke a écrit :
>> coroutine2 is a C++14-only library, providing cleaner code
>> (maintainability) + it fixes some issues (moveable-only parameters).
>> boost.coroutine will be deprecated.
> Up to when boost.coroutine will be deprecated? By what users of
> boost.coroutine in c++98 would replace the deprecated library?
> Is there any library in Boost that depends on Boost.Coroutine?

Boost.ASIO depends on Boost.Coroutine, although I didn't dig deep enough 
to know strong the dependency is.

> I've the impression that you will need to maintain it still for some years.

+1. As long as Boost.ASIO remains a C++03 library it cannot be ported to 
Boost.Coroutine2. Either Boost.ASIO has to be ported to a different 
coroutine implementation or Boost.Coroutine has to remain in Boost and 
be maintained.

>> I'm wondering if coroutine2 needs a review, even if the interface remains
>> the same compared to boost.coroutine.
>
> In your case, you are adding more but restricted to C++14 compilers. I
> don't think that review is needed in this is the case.
>
> I believe that it is up to you to decide if you needs a review. Spirit
> uses to request a review for each new version, but the interface are not
> preserved.

I don't think I've seen a review request for X3.

Anyway, Oliver, if your library design is the same and interface is 
unchanged then the only reason you might want people to review your 
library is to have someone inspect the implementation and perhaps docs. 
It's a very valid reason and you are welcome to request a review, but 
bear in mind that the main focus point of Boost reviews is design and 
interface, so you might not receive as much feedback. A mini-review 
might be better.

OTOH, if Boost.Coroutine2 is a new design then I think review is in order.