Tomorrow comes

haskell

Published on August 5, 2026

It so happens that in my case leaving a company where I spent the last 8 years sent me down memory lane. Among the effects of that little trip was a bit of nostalgia that made me remember particularly vividly the period roughly between 2016 and 2020, which I consider both very interesting and fruitful for the Haskell community and ecosystem in general, and for me personally. In retrospect, I must admit that this period was the peak of my Haskell-related activity. Later, my professional focus shifted to scalable build systems, Bazel in particular, while my personal interests moved to philosophy and painting in an attempt to attack questions that simply cannot be answered through software development.

Now, it’s been a while! Time sure flies. With a little bit of spare time on my hands, I decided to do a few useful things rather than simply remember and reminisce. This post presents them. It is a kind of public announcement, really.

My tutorials

One thing I am proud of is my Haskell tutorials. It is perhaps immodest of me, but re-reading them now I realize just how much work went into them, all well before AI. Back in the day if you wanted to write a good tutorial you’d read papers, PhD theses, and important blog posts (all of which you first needed to discover). You would do a lot of reading before you could write a single line of prose. The time investment there was quite substantial.

I hope that these tutorials can still be useful even in our day and age, and that understanding the subtleties of Haskell is still worthwhile. Acting on that belief, I have done three things to each of them:

  1. Fixed grammatical errors. Sadly, there were a few.
  2. Improved the flow and made the English prose tighter.
  3. Most importantly, updated all substance in order to align it with the current state of the ecosystem, as of GHC 9.14. In some cases, that meant writing entirely new sections (for example, on backtraces in the tutorial on exceptions).

I have also updated the list of recommended tutorials found on the same page. My goal there was to construct a reasonable, objective, and up-to-date selection of nice writing on Haskell. I will leave it to you to judge whether I have achieved that goal.

Ormolu

I created Ormolu in 2019 and have shaped its design ever since. My work on it continues at mrkkrp/ormolu, while an up-to-date Ormolu Live is now available at https://ormolu-live.markkarpov.com. Ormolu 0.8.2.0 is the first release made from the new repo and it is already on Hackage, where I continue to maintain the ormolu package.

To avoid confusion: the tweag/ormolu repository continues separately. If you get Ormolu through Hackage, Stackage, or the usual Haskell tooling, you do not need to do anything—the releases you receive are the ones I make.

I remain personally committed to maintaining Ormolu. Even though other things took a fair share of my mental space in the last 6 years or so, I have not abandoned a single practically useful Haskell package I have ever published—I assume a kind of responsibility over them. Come to think of it, Megaparsec is 11 years old at this point—it is a granddaddy of a package now! Some of my packages are more fun to maintain than others, but my idea of open source software development has always been about providing value and reliability rather than simple enjoyment of the process.

Next steps

I intend to resume posting here from now on. I will most certainly write about Bazel and about some open source work around it that I am doing right now. On the Haskell side of things, I am pleased with the state of most of my packages, and I probably won’t create new ones for now.

While working on this website’s code I did notice that one of my old projects—MMark—is actually not bad. It still powers this static website, and there are quite a few custom extensions that I wrote to adapt markdown to my needs. Using it, I get the feeling that it is built around a good idea. Unlike my other projects, though, it never received enough attention after an initial burst of development around 2018—other things got in the way. I am curious where I can take it now, at my own pace. My plan is to switch to a quote syntax that matches CommonMark and generally see how close to CommonMark I can get while still preserving what makes MMark itself: a strict processor where ambiguous input is an error rather than a guess. The idea here is to make MMark a little less surprising for people who are already familiar with markdown.

This is all I wanted to share for now. Enjoy summer, and Happy Hacking!