Ted, Keep in mind that I don't have any intent. I created this file by taking the Arrays.java from Apache Harmony and extending the sorting code in it to achieve the same results that the original Colt implementor got by extending the Sun JDK Arrays.java. For this particular function, I took the working 'int' copy, and modified it to take a swapper. So, the algorithmic intent belongs to whomever coded Arrays.java at Harmony.
The comments make sense to me, but I am really stuck on the conceptual question of how introducing a swapper breaks it. There has to be a place where the algorithm needs a value but has an index, I think. --benson On Fri, Jan 1, 2010 at 5:04 AM, Ted Dunning <[email protected]> wrote: > Generally the code seemed fine. > > Did you get a chance to look at my comments to see if I captured your > intent? > > On Thu, Dec 31, 2009 at 12:44 PM, Benson Margulies > <[email protected]>wrote: > >> I attached a patch to MAHOUT-235 that fails its unit test. I haven't >> spend \hours/ staring at it, but I have spent some time, and I feel >> like all this quicksort code is running together in my head. If >> someone is feeling charitable, or wants an opportunity to snicker at >> my expense ... >> > > > > -- > Ted Dunning, CTO > DeepDyve >
