Re: [boost] [review] Review of Outcome (ends Sun-28-May)

Subject: Re: [boost] [review] Review of Outcome (ends Sun-28-May)
From: Hartmut Kaiser (hartmut.kaiser_at_[hidden])
Date: 2017-05-26 13:53:23

> > On 05/26/2017 12:43 AM, charleyb123 . via Boost wrote:
> >
> >>       *- Thomas Heller -- (almost a review), ?reject, "not-ready-yet?"
> >>
> >
> > Please don't count my comments as a review. I don't feel qualified right
> > now to make any statements on the quality of the library presented.
> There
> > is too much in flux of it right now and after all discussions, it is not
> > clear to me what the current state of the library really is and how much
> it
> > will change after the review.
> 
> 
> You are making a valid point. Typically, the outcome of a boost review is
> etiher "accept" or "acceppt, but fix this and that", or "reject". Now,
> during this review, my feeling is we are just trying to design big parts
> of
> the library. There is so many open questions, that one cannot get te sense
> of the final shape. To reject it would be sending the wrong message that
> we
> do not wat the problem to be solved, or that the solution is wrong. But it
> is not wrong: it is just "I can make it whatever you want". To accept it
> on
> condition that it has a different interface, is like accepting something
> else than what we see. We can do it technically, but it feels wrong. I am
> confused here, I admit.

We have had such situations before, IIRC. The Boost Serialization library
for instance was rejected twice (!) before being accepted in the end because
the reviews ended up in a similar state (too many open design questions). 

Generally, I don't think that rejecting a library sends a wrong message. It
just means that it is not ready yet (for whatever reason, not necessarily
because of bad quality or similar). 

I for myself don't feel I'd be able to judge the library at this point as it
is not even clear anymore what we're reviewing... and I'm sure others feel
the same. But I would feel bad if the library got accepted now just because
people feel it would 'send a wrong message' otherwise. For this reason I
would like to suggest for the review manager to consider re-scheduling the
review to a date after the results of the discussion have been incorporated,
without making an acceptance decision right now.

Regards Hartmut
---------------
http://boost-spirit.com
http://stellar.cct.lsu.edu