> No doubt, but the actual function runtime is only one component of the > cost of applying it to a lot of dictionary entries --- I would think > that the table read costs are the larger component anyway.
Data domain can be not only dictionary but also something like article titles, urls and so on. On such relatively long strings (about 100 characters and more) this component will be significant (especially if most part of the table is lying in cache). In this case search of near strings can be accelerated in more than 10 times. I think that this use case justifies presence of separate leveshtein_less_equal function. ---- With best regards, Alexander Korotkov.
