[Issue 11597] Speed up std.random.dice

d-bugmail at puremagic.com d-bugmail at puremagic.com
Sat Nov 30 02:08:12 PST 2013


https://d.puremagic.com/issues/show_bug.cgi?id=11597


Joseph Rushton Wakeling <joseph.wakeling at webdrake.net> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |joseph.wakeling at webdrake.ne
                   |                            |t


--- Comment #3 from Joseph Rushton Wakeling <joseph.wakeling at webdrake.net> 2013-11-30 02:07:50 PST ---
(In reply to comment #2)
> Associated pull request:
> https://github.com/D-Programming-Language/phobos/pull/1716

Follow-up here based on feedback provided to Chris' pull request.  We agreed
that it wasn't an appropriate solution, simply because the cost of allocating
(ouch!) and then filling (O(N)) an array containing a cumulative sum of the
proportions, was actually worse than the (also O(N) but no allocation and no
extra operations) traversing of the proportions that dice() currently does.

However, it occurs to me that it might be possible to offer an overload for
dice() if it receives a SortedRange containing the cumulatively summed
proportions.  This could provide the best of all possible worlds as it might
offer a reliable means to binary-search dice _without requiring any allocation
at all_ (e.g. if your proportions are an iota(), or some other predictable
range).

-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------


More information about the Digitalmars-d-bugs mailing list