Re: [boost] Boost.Fit review Mars 3-20 result

Subject: Re: [boost] Boost.Fit review Mars 3-20 result
From: Rob Stewart (rstewart_at_[hidden])
Date: 2016-05-20 04:39:01

On May 19, 2016 7:58:51 PM EDT, Glen Fernandes <glen.fernandes_at_[hidden]> wrote:
>On Thu, May 19, 2016 at 7:07 PM, Vicente J. Botet Escriba
><vicente.botet_at_[hidden]> wrote:
>> Seen the reactions that I'm seen in this thread, I start to think
>that maybe
>> I could be wrong with my decision.
>>
>[snip]
>> I believe that we will need this 10-day review period if no more to
>review
>> the revised Fit library. This is why I'm talking of the need of a
>full
>> review. I don't consider to take time to review a library a bad thing
>as the
>> first goal is for me to improve the library in all its aspects.
>
>I did not observe the Fiber review, so I can't comment on that. I also
>agree with Andrey (and Vladimir and Robert)'s assessment: It seemed
>like your intention was a rejection (for this review) but that Paul
>was encouraged to address some of the issues raised, resubmit for a
>further review - which could carry a different result)
>
>This is probably because, as you've said, a full review of the library
>would be required.

Conditional acceptance has always meant that a library will be accepted once specific criteria, enumerated in the review summary, are met. A mini-review always followed to verify satisfaction of those criteria. If the library design, scope, or implementation must be subjected to further review, then the library is not acceptable, as is, for acceptance. In that case, it should be rejected.

Like Glen, I don't know (or recall) what happened with the Fiber review. The only mention I found immediately was Nat's summary[1]. If the Fiber review was not handled consistent with what I just described, we could revisit that, but note that there was one enthusiastic "yes" vote for that library (from Niall). I infer that Nat was, himself, in favor of the library once it satisfied the criteria he listed. The result was conditional acceptance and a mini review to follow. It seems justified, though the paucity of reviews casts a little doubt on it.

If you can produce a specific last of concrete criteria that Fit must meet such that it should be accepted automatically upon satisfying those criteria, then conditional acceptance is the appropriate result. If the list is extensive, but does not include open-ended changes, like redesigning a class hierarchy, sweeping API changes, etc., then the only question is how much time you think should be allotted to the mini review to be able to verify that the required changes have been made.

___
Rob

[1] http://lists.boost.org/boost-announce/2015/10/0456.php

(Sent from my portable computation engine)