fuzzy-find-test

Here is a little program for testing the performance of the fuzzyfind haskell lib.

The main logic is in https://github.com/aryairani/fuzzy-find-test/blob/main/src/Lib.hs

import Text.FuzzyFind
import Data.Foldable
import Data.List
import System.Environment

someFunc :: IO ()
someFunc = do
    queryArgs <- getArgs
    -- discard everything but the first field
    stdinLines <- fmap (takeWhile (/= ',')) . lines <$> getContents
    traverse_ (putStrLn . show) $ sort (fuzzyFind queryArgs stdinLines)

I compiled with -O2.

The shared.namespace.csv input I used is available at https://gist.github.com/aryairani/34d4b6cf65a7cb6117880c65d642a587. It was created with unisonweb/unison#2293 after pulling https://github.com/unisonweb/share into an empty unison codebase.

The command-line I used is:

stack exec -- time fuzzy-find-test-exe "M.tL" < shared.namespace.csv

which produced the top match contrib.bascott.continuations.v1._external.base.M1l.Map.toList in 2.5–3 seconds,

time fzf -n 1 -d, -q M.tL < shared.namespace.csv

which returned the top match base.Map.toList in 0.01–0.02 seconds

How can we improve it?