Over the weekend I attended the AmeriHac Haskell Hackathon. It was super fun and I learned a whole lot, and was exposed to even more. Everyone there was very interesting, helpful, and excited to talk to me. They all believe in the future of the language, and the only way that the language has a future is if people learn use it.

The event was hosted and sponsored by Jane Street, and they really outdid themselves to make sure we all had a pleasant time. Jane Street sponsors a lot of educational tools I like, like Advent of Code and some YouTube channels, so I have a very positive opinion of them. They want to hire people skilled in mathematics and computer science, so they support many free educational resources.

The main purpose of this hackathon was to bring together Haskell developers to work on the Haskell language compiler, tooling, and the package ecosystem. As I understand it, the main issue preventing Haskell from achieving mainstream popularity and industry adoption is not with the language itself. The bigger problem is that Haskell is unpopular, which leads to a lack of well-maintained packages.

I primarily sat with the people working on the dataframe Haskell project, a package that allows you to work with data in Haskell the same way you can in R or Python. I actually got my first real open-source contribution through this package, a single line that was basically dictated to me. Still, the experience was very fun. I use so many open-source tools and I really want to start contributing to them. The difficulty is finding an appropriate issue to fix and reading and understanding the existing code. Even going back to my old code and trying to comprehend it can be difficult, and understanding these massive projects with hundreds of interlinking files is remarkably difficult for me right now.

A lot of the discussion that went on was about how they can can help make Haskell more popular. One speaker who used Haskell almost exclusively for his startup lamented that although he loves Haskell and continues to use it, the lack of a well maintained library of packages makes it difficult to use at his company. He told a story how when their very first customer swiped their card, the payment didn’t work due to a bug in one of their dependencies. The overall unpopularity leads to a cycle where the lack of popularity leads to fewer packages, which in turns leads to even worse popularity. The people that love Haskell absolutely adore it, but it’s difficult to get new people in.

A common trait among all the people there is that they all value the elegance and logical beauty of their code. The speakers said that they care about the art and the craft of programming. I relate to this so much, and this is what originally drew me to Haskell in the first place. The way the syntax and language features come together allow you to write such elegant solutions to problems in a way that I find so satisfying. I love the list comprehensions, infix operators, and easy function composition.

Right now I am struggling to see where Haskell fits in. The people there (myself included) want to use Haskell for everything because it is the most fun language to use. But using Haskell is almost never an easy choice to justify since other languages just have better support that makes certain tasks nearly effortless. The dataframe support in Haskell is a huge, successful project, but Haskell will never get to the same ease and popularity for data work as Python or R. And why should it? I love the way data analysis pipelines work in R, and don’t really see a reason to use Haskell for the same task.

I’m trying to understand which domains are Haskell’s biggest strengths. Tiny little projects are easy to think about and don’t need the purity and explicitness of Haskell, but as projects get larger and more complex the power of the type system and pure functions becomes more apparent. Even in my ~3000 line Python project, the dynamic typing and unrestricted side effects already became a lot to manage. I can only imagine the compounding advantage of the type system in projects with hundreds of thousands of lines.

I am going to use Haskell by default as much as I can going forward. I really want to understand the ins and outs of the language, and the only way to get better is by deliberately choosing to use it. Coding in Haskell is just so fun, which is the most important thing long term. I love how Haskell lends itself to very elegant code and its connection to mathematics. Even if I never use it for a job, I want to use it for my personal projects. I want to read about the more advanced features of Haskell, like using the fancy language extensions like Template Haskell and type families. I’ve worked through hundreds of pages of Haskell books and solved tons of exercises, but these people at the conference were just on a whole other level that I want to get to.

Before this event, I had never had the chance to talk to other people that share these super niche interests in Haskell and category theory. Learning these things has been a completely solo endeavor for me, and it was so fun to talk to other people about it for the first time. It also really opened my mind about how much more there is for me to learn. I’ve already registered for ZuriHac which takes place this June, which would be so much fun (but also much more expensive for me).