Hacker News Reader

Comment by phreack | Hacker News Reader

I find this quite funny because it's my first time hearing I was supposed to be thinking of Elixir as a Ruby thing. I actually learnt about it from a concurrent computing class and it was always an Erlang thing, and now I know it as the magic sauce behind Discord that I always want to try and never find a good reason to.

View on HN

It's not really a weird thing because basically all of the Elixir/Phoenix core teams were big RoR contributors.

> I always want to try and never find a good reason to

As someone who learns best by doing, what are some practical projects that someone could do to learn Elixir? I know that Elixir is quite capable of solving certain kinds of problems very elegantly, but maybe my experience hasn’t presented these kinds of problems yet. Outside of building a Discord-like server or a Phoenix web app, what other good practical projects/applications are there for Elixir?

I'm probably the crazy one in the community who is using Elixir for the most super-strange things. For example:

- as a custom DHCP server to do multiple concurrent PXE booting (among other things)

- as a system for provisioning on-metal deployments (like ansible but less inscrutable).

- as a system for provisioning virtual machines over distributed datacenters.

I'll probably also wind up doing DNS and HTTP+websocket layer-7 load balancing too by the end of the year. Probably large-size (~> 1TB) broadband file transfer and maybe even object storage gateway by next year. I've rolled most of these things out to prod in just about year. I honestly can't imagine doing all of these things in Go, without a team of like 20.

Elixir sucks at:

- platform-dependent, like IOS, android, or like SDL games or something,

- number-crunchy, like a shoot em up, or HPC.

- something which requires mutable bitmaps (someone this past weekend brought up "minecraft server").

Actually even desktop might be okay, especially if you pair it up with electron.

> Elixir sucks at:

> number-crunchy, like a shoot em up, or HPC.

> something which requires mutable bitmaps (someone this past weekend brought up "minecraft server")

One thing I'd like to see for the BEAM communities long term are well maintained libraries of NIFs[0] for high performance and possibly mutable data structures. Projects like rusterl[1] and the advances made on dirty schedulers make this more feasible than it used to be.

It would be cool to write all the high level components of a minecraft-esque game in Elixir, and drop down to rust when you need raw performance. Similar to the relationship between lua/c++ in some modern game engines

[0] http://erlang.org/doc/man/erl_nif.html

[1] https://github.com/rusterlium/rustler

I'm in agreement; though I don't like rust (can't read it) and am the author of https://hexdocs.pm/zigler/Zigler.html

These are some good examples, and I highly appreciate you pointing out Elixir’s downsides.

What are you using to store state? The distributed system itself? An external database? Hybrid approach?

currently my main system in prod is a (very smart) stateless caching system, I don't have a database, but I will eventually spin up a (probably amazon) database, for the web gateway, when the keys to that are handed over to me. One of the nice things is that using elixir really puts you in the mindset of thinking hard about who should have responsibility for "the truth". No hiding behind objects backed by ORMs. State is pushed out as far as possible to the edge and I don't trust my server to keep state correctly. As soon as it's at risk of being inconsistent (netsplit, e.g.) the state-holders in the server are slaughtered, rebooted, and told to reconnect to the edge machines, which are trusted with the truth. The nice thing about elixir is that restart logic costs the the price of like 5-6 lines of code across the entire codebase (no worrying about cleaning up resources like tcp connections or anything).

Not in prod (personal project), I have a websocket-scraper on an aws free tier that's dropping data into a sqlite file, that I can transfer to my laptop for offline processing, so there are lots of interesting options.

This is really interesting to read, I’ve primarily thought about the DB as something which keeps your objects warm. Any recommended sources to read more about this way of treating state?

Haha I dunno, I just came up with it because that seemed like the sensible thing to do. And the virtualization library I use (libvirt) is basically a database anyways. Whenever I've talked with people who know better than I about distributed state (and iot) everyone seems to agree that pushing state management to the edge is the correct strategy in this case. For some things (like tracking customer intent and business logic) where consistency is important, like for you're responsible for someone's $$$, then you want to centralize and really be 100% certain your state is safe, and - I learned this analogy from stephen nunez, who got it from Toyota principles - you want a 'stop the world' lever when things go wrong. To prevent compounding inconsistency.

In crude and probably wrong terms, AP concerns can be more easily pushed to the edge and CP concerns like to be centralized.

When you say push state to the edge, would this be similar to how with the onion architecture you have the domain model in the center but repo access to get data on the outside?

Edit: So as not to have the DB be the ‘core’ of the app?

if you need to do some batching operations such as read from a queue, do some processing and publish to s3, I can recommend Broadway https://github.com/dashbitco/broadway

I have seen lots of codebases in lots of languages do this type of task, but aside from maybe spark on the high end, i haven't see it done better.

The beauty of erlang is your code reads like synchronous code so it's easy to read/maintain, but it has all the power of the parallelism of async code.

A BitTorrent client would be cool.

I wrote one in Scala+Akka and the actor model is great. It's even better when the runtime supports preemption, like Elixer.

An erlang BitTorrent client exists: https://github.com/jlouis/etorrent

Robotics/iot applications.

> I know it as the magic sauce behind Discord that I always want to try and never find a good reason to

Don't ever want to build a web app? That's pretty much Elixir's sweet spot, IMO. You'll get great productivity, great scalability, performance that's better than popular interpreted languages and a code base that's easy to reason about.

You'll also be able to do a lot from within the VM instead of relying on external services.

In my mind, in a web context, Elixir is a better Nodejs.

It will never be as popular though.

And it's not like other languages stay stagnant. JS and TS, Java with green threads, C# with actor and LiveView like libraries.

Aside from getting rid of decades of language cruft, mutability, etc, one thing I doubt Node will ever have is a fairly unified ecosystem with a dominant framework like Phoenix.

I migrated the code base for my last startup from Node to Rails at the end of 2015, then from Rails to Phoenix at the beginning of 2016. Pretty much all of that Elixir + Phoenix 1.1 code is still fine today. There haven't been any breaking language changes and the most arduous part of upgrading Phoenix to the current version would be the front-end—replacing Brunch with Webpack and removing Bootstrap. Other than that it's just a few lines of code.

The Rails app is slightly more dated but still fairly trivial to update. Most of the gems still do things in the same way and there's generally a clear way to go. The JS app on the other hand, is a pretty depressing mess of broken dependencies and multiple libraries I was using have dramatically changed their APIs.

For me, productivity and ease of maintenance matter much more than popularity does when picking a tech stack.

This strikes me as someone chasing the shiny.

Why migrate working code to a new language?

And to do it twice?

The first time was early on while still at an exploratory stage. I made the change for productivity reasons. Rails did pay off on that axis, even though I was much less experienced with it than the original stack.

The second time was because I'd gotten it to where it was getting users but I couldn't afford the server costs (given that it was free to use and my users were largely in Vietnam and Thailand). I needed something that could scale much more cheaply.

If I'd had millions of dollars in the bank, I'd have just stuck with what I knew and added people and servers as necessary.

Same. I have never even heard there is a relation to ruby, I always just thought it was just a functional language for the Erlang runtime. I had no idea the syntax was "Ruby like".

Discord switched to Rust some time ago: https://blog.discord.com/using-rust-to-scale-elixir-for-11-m...

Still a very impressive use of elixir IMHO.

The article says nothing about them switching to rust. They implemented a specific data structure in rust using elixir NIFs (native interface). This is equivalent of calling a C library from other languages.

"Our core services aren’t written in Rust, they are Elixir-based."

They haven't switched, just use Rust at some places for performance improvements.

My bad, I visibly remembered the article incorrectly and didn’t take the time to read again when sharing.

Thanks for the correction. Unfortunately I cannot edit my original comment.

Others have already repeatedly mentioned this was a performance optimization. Elixir/Erlang is not particularly good for heavily CPU bound tasks--This is by design. For lack of a better source, here is a post on the Elixir forum from Robert Virding (one of the creators of Erlang) on the subject: https://elixirforum.com/t/on-why-elixir/34038/62

Based on article they didn’t switch to Elixir but instead are using Rust within Elixir in order to solve complex performance issues.

Reply on news.ycombinator.com