Re: [boost] [xint] Boost.XInt formal review

Subject: Re: [boost] [xint] Boost.XInt formal review
From: Jeffrey Lee Hellrung, Jr. (jhellrung_at_[hidden])
Date: 2011-03-03 13:42:28

On 3/3/2011 10:30 AM, Christopher Jefferson wrote:
>
> On 3 Mar 2011, at 17:07, Jeffrey Lee Hellrung, Jr. wrote:
>>
>> You can't assume that everyone will be satisfied with your integer_t data structure.  Someone may want a data structure that uses the small-object optimization, because they have a lot of tiny integers running around and occasionally have large ones.  Or they may want to use a std::deque(-like) underlying storage.  I don't know.  But I do know that your library will be strictly more usable if you decouple the algorithms from the integer_t data structure.
>>
>> I'm hesitant to vote for acceptance unless this decoupling were effected; your mileage with other boosters will obviously vary, but I gather Mathias is of the same opinion.
>
> My biggest concern with decoupling is that introduces a much larger public API. At the moment it would be possible to make very large changes to the internals of xint without changing the public interface. Also, without multiple different data structures to store the internal data, it is not clear what a clean decoupling would look like.
>
> Decoupling would make evolution of the library much more difficult, whereas decoupling later would be easier.

You're right, those are valid concerns against decoupling, and in 
retrospect I should've clarified that "hesitant" did not necessarily 
mean "opposed".  At the very least, I would like to see the extraction 
of the algorithms independent of the data structures seriously considered.

- Jeff