Re: FW: Join based upon LIKE

[(Hal]

>>>> 2011/04/28 15:28 -0400, Jerry Schwartz >>>>
No takers?
<<<<<<<<
And this is not real taking, because the algorithm of which I am thinking, the
edit-distance (Levens(h)tein-distance) algorithm costs too much for you (see the
Wikipedia entry). The obvious implementation takes as many steps as the product of the
two compared strings s length. On the other hand, a good implementation of "LIKE" costs
the pattern s length added to all the strings against which it matches s length, a sum,
not product, of lengths.