
On Tue, Nov 13, 2007 at 04:33:51PM +0800, Ben Bear wrote:
2007/11/13, Hervé Brönnimann <hervebronnimann@mac.com>:
coherent interfaces. Please look at:
I'll read this proposal. It's a little long for me.
I did so already and found two minor issue: To get all (n,r) combinations I have to specify r as value "middle". According to the document it is possible to specify r=0 but not r=n? ("Without repetitions, r is specified by a middle position in the input range [first, last) which is r positions away from first.") If r=n is possible, middle=last would be a valid choice ==> [first, last)=>[first, last]. I also do not understand "Permutations and combinations can be ordered lexicographically, starting with the subset at the first k positions of the sorted total range, and ending with the subset at the last k positions of the same range. Thus the effects of the algorithms are completely and unambiguously defined." What can be sorted? The elements of each r-tuple or do you want to say that all determined permutations and combinations are sorted (provided in a special order)? Probably you want to use r instead of k in the text above? I first thougth that I can get all values in a single r-tuple unsorted by specifying a parameter k=0 and wondered that it affects the beginning and end of some data. Could you please try to improve it a little bit? Maybe by giving some examples in the documentation? Thanks, Jens