Talk:Sorting algorithms/Tree sort on a linked list

Revision as of 10:55, 11 May 2014 by rosettacode>Dkf (Bah, the task is even worse than that as I look more at it. Delete?)

I don't understand this task. Tree sort traditionally means loading the elements into a (new) binary search tree data structure, and then reading the elements back out. I don't understand what it means to "tree sort them inplace" in a linked list. --Spoon! (talk) 03:08, 4 April 2014 (UTC)

The proposed task talks about performance. This is a Bad Idea because it's next to impossible to compare performance between systems (different CPU speeds, different memory bandwidths, different loading patterns, etc.) Talking about performance strongly encourages people to try to “optimise” their implementations, which tends to make them significantly less readable and less idiomatic. Finally, actually measuring performance fairly and accurately is hard; there are lies, damned lies, and benchmarks. –Donal Fellows (talk) 10:49, 11 May 2014 (UTC)

And the proposed sample text is ridiculously long; there's also no meaningful output. (Just printing some metadata about the supposed performance is insufficient, as it does not check that things are correct. Wrong “solutions” can be incredibly fast.) This whole task needs major revision (and soon!) or I'll have to consider deleting it entirely. –Donal Fellows (talk) 10:55, 11 May 2014 (UTC)

Return to "Sorting algorithms/Tree sort on a linked list" page.