中值排序基数法（Median sorting Radix method）.doc

中值排序基数法（Median sorting Radix method） Implementation of tree structure with median sorting Radix method In the preparation of BBS, people have often asked how to implement tree structure An irresponsible answer is to use recursive algorithms. Recursion is, of course, a feasible solution (the two fork tree movesthrough seemed to only use the recursive algorithm), but for BBS, it was a large number of Sql query (although you can use stored procedures to do, but fundamentally to speed up, you should consider a faster algorithm). Here is an algorithm to implement a tree structure that is feasible and completely discarded. Here is another way to implement a tree structure using the median sorting Radix method: First, the main idea: to add a sort of Radix field ordernum, reply to the same post, insert the post, sort Radix ordernum, take the median of the two. For the brevity of the narrative, the fields related to the tree structure are discussed only here. Add three redundant fields in the table, rootid -- for recording root ID, deep - for recording the depth of the reply (representing root paste at 0), and ordernum -- sort base (key). Table forum and (only columns related to the tree structure):id, rootid, deep, ordernum, in which ID, rootid, and deep are int (deep is tinyint) and ordernum is float. (in this example: for simplicity, using a small initial order base, in the practical application, should be used to start large base, and should take an integer power of 2, such as 65536=2^16, what sort are sorted by ordernum value). ID rootid deep ordernum 1000 21164 ______________________________ 31132 replies first pastes, takes 1 and 2 cardinal number median, namely (0+64) /2 After sorting, the result is: ID rootid deep ordernum 1000 31132 21164 ______________________________ 41248 replies third pastes, takes 3, 2 base median value, namely (32+64) /2 After sorting, the result is: ID rootid deep ordernum 1000 31132 41248 21164 ______________________________ 51356 replies f