From: Hervé Brönnimann (hervebronnimann_at_[hidden])
Date: 2008-02-05 23:30:03
Hi Ben: First of all, we missed (knowingly) the deadline for the
Bellevue meeting. I am simply too busy to finish the proposal
although I thought it was pretty much over before I saw your comments.
About the count functions: I'm not sure I like the non-count
functions anyway, and I don't think it's reasonable to provide the
transform functions in a C++ standard proposal. This is not
essential to C++ std lib, so it wouldn't have any support if it's too
big / too much at the fringe. So if we don't like the count
functions as they are, then I'd say let's take them out of the
proposal and only propose the next/prev_partial_permutation and next/
prev_combination functions. Makes the proposal more focused.
Nevertheless, a boost submission would have count and transform
functions, that would be no problem.
-- Hervé Brönnimann hervebronnimann_at_[hidden] On Feb 2, 2008, at 9:22 PM, Ben Bear wrote: > 2007/12/26, Hervé Brönnimann <hervebronnimann_at_[hidden]>: >> Regarding the ordering of the sequences for combination counts, your >> note was very interesting. I don't know which one is more valuable, >> the (ordered) sequence of combination counts, or the sequence of >> counts such that the actual combinations are in lexicographical >> order. Let's make a decision, soon. One important factor (for me): >> Is there an implementation of your algorithm (sequence of counts such >> that the actual combinations are in lexicographical order) as a free >> function, without state? I.e., with the interface that I give in the >> proposal? >> > > > About Repetition Combination Interface > > The Repetition Combination is useful, even for testing gacap itself. > > > I think that the count versions have some problems: > > 1. The interfaces are inconsistent with other functions. > 2. A transform function is needed for converting the numbers to real > values. If we don't provide this, it will be not easy to use. > 3. There's two types of the numbers' lexicographic order, the > numbers' or the real values'. > > > The count versions are beautiful, but they are not a good choice for > this proposal. Users must spend more time to learn those count > functions. > > ... > > Herve: I think I lost the steps... Shall you start a new thread for > the proposal? This may help the discussion. > _______________________________________________ > Unsubscribe & other changes: http://lists.boost.org/mailman/ > listinfo.cgi/boost
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk