Re: async-mqtt5 is CONDITIONALLY ACCEPTED

From: Ivica Siladic (ivica.siladic_at_[hidden])
Date: 2024-10-30 12:23:11

Hi everyone,

On behalf of the Mireo team, I’d like to express our gratitude to the Boost community for the inclusion of Async.MQTT5 (or even better boost.mqtt5) into Boost. Special thanks to Klemens for managing the review process, and to all the reviewers for their valuable comments, questions, and detailed feedback!

Async.MQTT5 was designed to be extremely easy to use while still allowing for some customization. Our goal was to abstract away most MQTT protocol details and network management to the greatest extent possible, making it accessible to developers with minimal MQTT knowledge. Naturally, this approach may make the library less suitable for advanced use cases where complete control over the protocol is required. However, we hope it will serve a wide range of use cases where projects can benefit from a plug-and-play component that “just works.” Having the library included in Boost certainly strengthens the confidence of developers looking to rely on it.

Thank you once again,

Ivica Siladic

> On 30.10.2024., at 09:45, Klemens Morgenstern via Boost <boost_at_[hidden]> wrote:
> 
> On Wed, Oct 30, 2024 at 4:04 PM Thomas Fowlery
> <thomas.fowlery.yes_at_[hidden]> wrote:
>> 
>> I worry about the integrity of the Boost review process. The review manager and all the accepting reviewers were/are affiliated with the C++ Alliance.
> 
> That is incorrect, Marcelo is not and has never been associated with
> the C++ Alliance. I would recommend getting facts like this correct
> when discussing integrity.
> 
>> The review manager thanks these reviewers, but not the ones who rejected.
> 
> I thanked all the reviewers, but pointed out those who invested more
> than 8h. The rejecting reviewers did not state how much time they
> invested. I did state how much I appreciated Daryll's review and all
> the detail he went into.
> 
>> The outcome of the review is presented as a foregone conclusion, with no analysis of the actual topics that have been discussed.
> 
> The outcome of the review is the conclusion of the review.
> 
>> The review manager does not state how the different reviews and issues have weighed in on his decision. There is at least one accepting review where the reviewer hasn't even built and ran a single example. Meanwhile, the two rejections (and some non-review comments) brought up major design questions which haven't been addressed by either the review manager or the author(s). These facts make the outcome way less clear, in my opinion.
> 
> That's a fair point, so let me add to this: The two rejections were
> essentially based on the fact that async-mqtt5 is an asio-only library
> and too high level. In either case the reviewers would want a
> different kind of library than under review.
> Four reviewers thought the basic idea of the library (a high level
> client based on asio) was the right idea, so that's a 4:2 - I wouldn't
> know how to weigh a basic preference like this.
> 
> Otherwise I put the most weight on Marcelo's & Ruben's reviews on the
> Accept side, as they are authors of asio-based libraries, have
> invested a fair amount of time & pointed out some significant details
> (showing they really looked into it).
> On the Reject side I weighted Daryll's review most highly, as he's
> experience with both asio & mqtt and his review gave the impression
> that he really looked deeply into the library.
> 
> I hope that helps to alleviate your worries.
> 
> _______________________________________________
> Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost