3 Jun
2009
3 Jun
'09
6:02 a.m.
On Tue, Jun 2, 2009 at 11:42 PM, Scott McMurray <me22.ca+boost@gmail.com> wrote:
... but I'd claim radix sorting is a common enough goal that a nicely-tuned STL-style implementation would be appropriate for boost.
Radix sorting is indeed appropriate, IMO. However, there are many "proven" radix sort algorithms, and (probably) many more "unproven" algorithms. Seems to me that boost is a set of "generally useful" libraries, and not a "proving ground" for new algorithms. The point in question is then: At what point is an algorithm adequately "proven" for inclusion in a boost library? Jon