tl;dr - I wrote a small library for CountMin sketches in Haskell (you can find the code on GitLab, @ mrman/haskell-countmin). I then try to optimize it and fail, but throw in an example of doctest usage, CI with Gitlab, and how to host the generated Haddock on GitLab Pages. In some work and exploration earlier this year I found some time to read more about sketches – probablistic data structures usually used for summarizing data.
