[boost] [sort] Re: [review] Formal review period for Sort library begins today, November 10, and ends Wednesday, November 19

Subject: [boost] [sort] Re: [review] Formal review period for Sort library begins today, November 10, and ends Wednesday, November 19
From: Vladimir Prus (vladimir_at_[hidden])
Date: 2014-11-10 02:35:06

On 11/10/2014 08:37 AM, Edward Diener wrote:
> The formal review of the Sort library by Steven Ross starts today, November 10 and is scheduled to continue through November 19th.
>
> About the Sort library
> ==================
>
> The Sort library is a library which implements hybrid sorting algorithms
> based on a Spreadsort ( http://en.wikipedia.org/wiki/Spreadsort ), of which the author of the library, Steven Ross, is the inventor.

Apologies if I'm overly critical, but we're asked to review an implementation of an algorithm from Steven's paper from
2002, and as of today, neither Google Scholar nor Citeseer can find any references to that paper. So, it's not an
established algorithm by any means, Boost is generally not an algorithms review organization,  and so I'm not sure it's
good for Boost to review the proposed implementation at all.

>       - What is your evaluation of the potential usefulness of the
>         library?

I am not sure the library is useful. For very few users, sorting 100kb arrays of integers is a performance
bottleneck. And if it's truly a performance bottleneck, one would probably ponder cache architecture, or use of GPU, or
a zillion other tricks. Like this paper from 2010 appears to do, for example:

        http://dl.acm.org/citation.cfm?doid=1807167.1807207

> And finally, every review should attempt to answer this question:
>
>       - Do you think the library should be accepted as a Boost library?

I think the library should not be accepted. There's no reason to give Boost approval to implementation of algorithm that
is neither accepted classic nor an obvious breakthrough.

-- 
Vladimir Prus
CodeSourcery / Mentor Embedded
http://vladimirprus.com