Re: async-mqtt5 is CONDITIONALLY ACCEPTED

From: Peter Dimov (pdimov_at_[hidden])
Date: 2024-10-30 17:10:45

Andrey Semashev wrote:
> On 10/30/24 17:14, Vinnie Falco via Boost wrote:
> > On Wed, Oct 30, 2024 at 7:01 AM Christian Mazakas via Boost <
> > boost_at_[hidden]> wrote:
> >
> >> The author of the proposed library avoided all discussions about the
> >> design
> >
> > There was quite a lot of discussion regarding the design, and it took
> > place on Slack rather than the mailing list.
> 
> I think, the formal media for the review discussion is the mailing list.
> Yes, reviews may be collected through other channels, but the discussion
> needs to be held in one place that is easily referenceable, and this is currently
> the ML.

I agree with Andrey. Design discussions about a library currently under
review are supposed to take place on the mailing list. This serves two
purposes, it helps reviewers who for any reason are struggling to reach
a final accept/reject verdict, and it can encourage people to review the
library by piquing their interest.

In addition, having the discussion archived is both of historical interest
and helps the review manager write up a summary. Should he deign to
do so, of course.

Going back to the original question of whether the library author is
expected to answer questions about the library under review: in short,
yes.

This helps reviewers evaluate the soundness of the design decisions
and get a feel about how the library has evolved.

E.g. a potential reviewer might ask "I see that you're doing X, but have
you considered doing W instead?" and the author would reply "yes,
we used to do W, but this turned out to not be a good idea in practice
because of reasons A and B, and because users tended to fall into trap
C, which X avoids."