In the last weeks Lemmy has seen a lot of growth, with thousands of new users. To welcome them we are holding this AMA to answer questions from the community. You can ask about the beginnings of Lemmy, how we see the future of Lemmy, our long-term goals, what makes Lemmy different from Reddit, about internet and social media in general, as well as personal questions.
We’d also like to hear your overall feedback on Lemmy: What are its greatest strengths and weaknesses? How would you improve it? What’s something you wish it had? What can our community do to ensure that we keep pulling users away from US tech companies, and into the fediverse?
Lemmy and Reddit may look similar at first glance, but there is a major difference. While Reddit is a corporation with thousands of employees and billionaire investors, Lemmy is nothing but an open source project run by volunteers. It was started in 2019 by @dessalines and @nutomic, turning into a fulltime job since 2020. For our income we are dependent on your donations, so please contribute if you can. We’d like to be able to add more full-time contributors to our co-op.
We will start answering questions from tomorrow (Wednesday). Besides @dessalines and @nutomic, other Lemmy contributors may also chime in to answer questions:
Here are our previous AMAs for those interested.
What have been the biggest challenges with the project over the years, both in terms of technical and non technical aspects. I’d be interesting to hear a bit of retrospective on how has the stack’s been working out, and what surprises you might’ve run into in terms of scaling and federation. What recommendations you’d make based on that and what you would’ve done differently knowing what you know now.
The stack is great, I wouldnt want to change anything. Postgres is very mature and performant, with a high focus on correctness. It can sometimes be difficult to optimize queries, but there are wizards like @dullbananas@lemmy.ca who know how to do that. Anyway there is no better alternative that I know of. Rust is also great, just like Postgres it is very performant and has a focus on correctness. Unlike most programming languages it is almost impossible to get any runtime crashes, which is very valuable for a webservice.
The high performance means that less hardware is required to host a given number of users, compared to something like NodeJS or PHP. For example when kbin.social was popular, I remember it had to run on multiple beefy servers. Meanwhile lemmy.ml is still running on a single dedicated server, with much more active users. Or Mastodon having to handle incoming federation activities in background tasks which makes the code more complicated, while Lemmy can process them directly in the HTTP handler.
Nevertheless, scaling for more users always has its surprises. I remember very early in development, Lemmy wasnt able to handle more than a dozen requests per second. Turns out we only used a single database connection instead of a connection pool, so each db query was running after that last one was finished, which of course is very slow. It seems obvious in retrospect, but you never notice this problem until there are a dozen or so users active at the same time.
With the Reddit migration two years ago a lot of performance problems came up, as active users on Lemmy suddenly grew around 70 times. You can see some of that in the 0.18.x release announcements. One part of the solution was to add missing database indexes. Another was to remove websocket support, which was keeping a connection open for each user. That works fine with 100 users, but completely breaks down with 1000 or more.
After all there is nothing I would do different really. It would have been good to know about these scaling problems earlier, but thats impossible. In fact for my project Ibis (federated wiki) Im using the exact same architecture as Lemmy.
It’s great to hear things mostly worked out. Stuff like scaling bottlenecks is definitely tricky to catch until you have serious loads on the system, but sounds like the fixes very mostly trivial validating overall design. It also looks like you managed to get a way with a fairly simple stack by leveraging Postgres and Rust. I’ve had really good experience with using pg myself, and really don’t see a point in using anything else now. You can use it both as a relation db and a document store, so it’s extremely flexible on top of being highly performant. Keeping the stack simple tends to be underappreciated, and projects often just keep adding moving pieces which end up adding a lot of overhead and complexity in the end.
2nding @nutomic, that I’m really happy with the stack.
The one that seems really magical to me, is diesel. With it we get a compile-time-checked database, that’s tightly integrated to the rust objects / code.
Every single join, select, insert, etc is checked before lemmy is even run, and it eliminates a whole category of errors resulting from mismaps.
Its made adding columns, and changing our data structures so much less error-prone than when I lived in the java-world.
Whenever we find that we’d want to do things differently, we usually do a refactor ASAP so as not to keep rolling spaghetti code. We’ve had to do this many times for the federation and DB code, and even have 2 major refactors that also add features ongoing. But luckily we’ve been able to stay in the rust eco-system for that.
As for UI, leptos didn’t exist when I built lemmy-ui, so I went with a fast react-like alternative, inferno. Its showing its age now, so @sleepless1917 is working on lemmy-ui-leptos, which hopefully will supercede lemmy-ui.
I briefly worked with Hasura which does this sort of magic to produce GraphQL API on top of Postgres. Incidentally, also written in Rust. I do like Leptos approach, it sounds similar to HTMX approach where you just treat the DOM as a dumb terminal.
The stack is overall amazing, but not perfect. Waiting for the Rust code to compile is sometimes very annoying, but I wouldn’t want to use a different language. And we had to implement somewhat complicated things that existing Rust libraries did not do. For example, I made the “i_love_jesus” library so Lemmy could have cursor pagination that uses indexes well and allows bringing back the “back” button, we have a few custom QueryFragment impls because of diesel’s limitations, and we have a custom migration runner to do fancy stuff (see crates/db_schema/src/schema_setup.rs).
Could I ask if there’s any meaning behind that name?
Nothing related to the library. I love Jesus. I’m Catholic. I’m a little silly. Also my GitHub profile picture is Jesus.
I don’t know what I expected lol. Cheers.
Edit: I don’t mean this disrespectfully, it was just a very direct and obvious answer.
Being a Lisper, the idea of waiting for the compiler is very jarring. :)
Oh, to be able to develop Lemmy with something like SLIME or Geiser, now that would be a dream. Too bad the CL’s library ecosystem is so much worse than Rust’s.
Clojure has a lot better story in that regard living on the JVM, but the overhead of using the JVM is a downside of its own. It’s a good platform, but definitely not what you’d call lightweight.
Yeah, if I was building something production ready in Lisp, Clojure would be my choice even though I prefer CL. Ecosystem is ultimately king.
For sure, having mature and battle tested libraries is really hard to beat.