The hard work

Look at anything that has achieved great outcomes. Worker's movements, liberation movements, open source projects, great works of art or even solid software. Often there is a simple explanation of how it works or came to be. The truth of how it happens is always a mess of humans, hard work, uncertainty and eventually ambiguous results, step by step, that are only recognized as greatness and a whole much later. We appreciate the first protest because it is a step towards the greater outcome. We appreciate the artist's older work because it shows the path to what got them recognized. We love that Debian and Linux exist but only the people involved see the complex human endeavour that makes it happen.

The complex and nuanced humanity of things, the progression from origin to achievement and the circumstance in which something exists are all important. Even in technical decisions the nuances are really what it is all about. When a developer says "It depends" it is a way to acknowledge the complexity. If they then bail out and won't explore the options they aren't really doing the work. If they then start weighing the trade-offs they are forging a path, which is the work.

Explaining to people why I like Elixir is something I've gotten a fair bit of practice at just due to talking about it a lot. I still don't like my answers. One of the snappiest ones is that "it was built for creating services, now everything is services". But that's an explanation of Erlang. So now you have to get in to the nuances and complexity of the story.

A thing that doesn't matter or will be considered a detriment according to some people is that Elixir is not owned or driven by a megacorp. Most of the development is by Dashbit. A small business specialized in Elixir, created by the original creator of Elixir, to support Elixir. Erlang is a corporate language but not in the way of Go. Ericsson is a large company, not a FAANG by any stretch though. The Erlang part is of Ericsson punches way above their weight class in terms of what they do. But they are tiny. They do get contributions from Meta who uses it heavily in WhatsApp. It is clearly not a Meta project though. Organizationally Elixir is independent. There is the Erlang Ecosystem Foundation which is joining the Linux Foundation. The foundation is currently mostly a diplomatic entity for the big picture, it doesn't drive development.

The community is a bit complex in the split between Erlang and Elixir along with growing interest in Gleam. It muddles the message to talk about this whole ecosystem. The communities cross-polinate and benefit each other as well. And it would be a dick move to not collaborate. If you think the community is entirely cohesive across Erlang and Elixir though you'll be surprised. Some people move all around with no issue but there are definitely large cultural differences between Erlang and Elixir that provide quite some spice to the mix.

On the technical side Elixir is just an absolute mess of pragmatic balances and trade-offs. This loses the people who want conceptual purity. You don't get statically typed FP full of monads. You don't get any OOP. You don't get absolutely pure functions. But there is a degree of control of side-effects. Erlang struck the foundational balance. Utilitarian functional programming in service of concurrency, resiliency and code updates.

On the syntax and tooling side it has the Ruby heritage which focuses on being approachable, easy, friendly. But differently from Ruby it has the incredible technical depth and wild capabilities of Erlang/OTP. It is not the most beginner-friendly. Partially that's a limitation of not having enough developers working on it. Partially it is because there is a tension between beginner-friendly and capable in design.

There is often no point in teaching someone lessons about resilience and scaling before they have experienced some degree of pain in that area. This is why it doesn't matter that Python is slow. It doesn't matter how weird Node.js is a bout multiple cores. You need some scars before you start looking at concurrency and considering Go, Elixir and .. Rust?

There are small choices, like Oban vs Broadway vs something custom. Or between using Phoenix LiveView or a full front-end framework. While we can usually make good default recommendations for the 80% case the reality is full of nuance. Every time.

We have a ton of cool answers to really hard and frustrating problems in Elixir. Some of the things we get by default would be massively painful to achieve in other ecosystems. On the flip side we have a smaller ecosystem of ready-to-go libraries, smaller knowledge base, more things you need to learn to use it well. And of course no-one ever got fired for choosing Go/Python/TypeScript/C#/Kotlin. They were just doing what everyone else did.

And we are quirky as a community. Because we have different possibilities we are generally not as inclined to just grab the same off-the-shelf solution everyone else does because we don't need as much of it. Or we see a potentially better way of doing things with our primitives, our building blocks. Of course that means we set ourselves a bit apart. This might be something you like or something that frustrates you. It can be both.

Pitching Nerves to someone is an absolute minefield of "it depends". Are they aware of the difference between microcontrollers and larger embedded devices? Are they on the Raspberry Pi-curious end of the spectrum or more industrial PC developer side? What type of thing are they looking to do? How comfortable are they with the Linux parts? How comfortable are they with Elixir? This tells me where they may face difficulty and where things will be straigthforward. This is not a flaw in Nerves. It is just a complex topic and helping people find a good entrypoint is a decent bit of work.

I really appreciate that Elixir is independent. I really adore that we take a second look at problems and don't just do the default. I like that the community is a size where I can at least recognize many of the names, though that will pass. By now I'm very comfortable with our various building blocks. Being without trivially creating actors, doing message passing and handling concurrency would be such a pain. I am used to the cultural landscape. I have a sense of how the community operates. I have friends, peers and colleagues, a sense of community in Elixir and the BEAM ecosystem. Transmitting the various factors that make it great for me is not a snappy one-liner. It is about where it came from, what it is good for, what it chooses to not be, who is involved, how they think, how they do things and it is about how the whole fits with my desires.

If you do Elixir, why? If you don't, why? And what is your favorite human mess recently? Let me know on the Fediverse where I'm @lawik@fosstodon.org or by responding to this email to lars@underjord.io.

Thank you for reading. I appreciate you.

The Elixir shirt pre-orders will ship sooooooon oswag.org.