mudderjs VS StringDistances.jl

I have solved this problem for my own purposes — no claims of grand scalability or high efficiency — by storing indices or "ranks" of items as strings (Postgres TEXT) using a library called mudderjs[0] and a thin wrapper around it[1]. Sorted lexicographically (in dictionary order) arbitrary-length strings have arbitrary precision. You can always find a string between any two strings; for instance, between "a" and "b" is "am" and between "a" and "ab" is "aam". You do have to have the entire ordered collection in scope to generate a new rank for an item, but reordering an item only requires updating one row and isn't subject to floating point precision.

[0] https://github.com/fasiha/mudderjs

[1] https://github.com/pubpub/pubpub/blob/master/utils/rank.ts

You'll need: * StringDistances.jl to calculate the distance between the strings. * FlexiJoins.jl to join the dataframes based on a predicate. * A custom function that you write that takes two strings as arguments and returns a boolean for if the strings are "close enough" based on their distance.

If you need to know exactly what the diff is, you might want to use something like github.com/google/diff-match-patch. Otherwise, a simple Levenshtein distance would suffice. This library seems to have a whole bunch of string distances implemented. Hope this helps!