[boost] Re: Typeof: Review result

From: David Abrahams (dave_at_[hidden])
Date: 2005-06-07 20:37:47

"Andreas Huber" <ahd6974-spamgroupstrap_at_[hidden]> writes:

>>>> I ended up getting into an unpleasant exchange with the library
>>>> author, who repeatedly challenged me to suggest concrete changes to
>>>> the design to fix the perceived problems. I had examined the library
>>>> more thoroughly, my guess is that I would have been able to suggest
>>>> improvements. I don't blame the library author in this case; it's
>>>> only natural to ask for an alternate design when you are told that
>>>> your design is flawed;
>>
>> Maybe, but you shouldn't feel guilty.  The onus is on the proposer to
>> come up with a good design.
>
> The question is: How far does a library author need to go in providing 
> evidence that the design is "good" (which often means different things 
> to different people, but lets ignore that for the moment)? Does a 
> proposer need to "prove" that the library design is the best currently 
> imaginable? 

A proposer needs to make a judgement call about which objections are
worth trying to satisfy.  Sorry, there aren't any hard and fast rules
here.

> While this might be possible for some libraries I don't think it is
> generally feasible. More specifically, if a raised point is so vague
> that the library author is at a complete loss exactly how an
> improvement could be implemented I think it is only fair to turn the
> roles around and require the reviewer to at least outline how the
> improvement is implementable within the given requirements.

You can't "require" anything of the reviewer.  It's okay to ask, of
course.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com