Thanks for the suggestions!

Create an ARCHITECTURE.md document explaining the overall architecture and different components of Lemmy. This would help new contributors quickly familiarize themselves with the codebase.

The documentation has some info about the architecture as well as how to start contributing. Is there anything particular missing, or do people not find these docs?

Publish a public roadmap with milestones and release plans. This gets people excited about the project’s direction and motivates them to contribute.

As mentioned in the other comment, we will have a public roadmap once the NLnet milestones are finalized. Other than that the issues are up for grabs for anyone, so Im not sure what sense it would make to give them priorities.

Nightly builds enable contributors to test upcoming changes, offer feedback, stay actively involved in the project’s progress, iterate quickly on improvements, and foster a culture of continuous improvement.

These are available as :dev Docker images, and deployed to test servers at enterprise.lemmy.ml etc. Maybe this should be documented somewhere to let people know?

Implement a bounty system for critical or challenging issues to incentivize contributions. Some people who don’t want to contribute with a monthly subscription may prefer this kind of contribution.

Bounty systems dont work well in my experience. Usually the reward is too low to pay an actual dev salary, and then its still a hassle to get the money into your bank account. Plus there is no gurantee to get paid for your work, its always possible that someone else submits a solution just before you finish yours.

Create project swag (stickers, t-shirts, etc.) and distribute it to active contributors as a token of appreciation.

Good idea, though we dont really have the funds to pay for that. It was also suggested to sell swag to earn money for Lemmy, but havent gotten around to that yet. Do you by chance any good company to produce and sell this?

Institute a “Lemmy contributor mentorship or apprenticeship” program where experienced developers formally take up 1-2 promising new contributors under their wing.

Well if anyone opens a pull request we give feedback as part of the review process. And if anyone has questions we also answer them in the issue tracker or on Matrix. But in practice there seems to be little demand in this regard.

Live or recorded screencasts solving issues, similar to mentorships but instead of one-on-one it allows more people to feel engaged in the development process, and provide feedback in the case of live streams.

That seems to get more into entertainment, not what we want to do.

Host discussions about issues on Lemmy itself, as suggested in the blog post below. The voting and threads with nested comments make it easier to have productive conversations compared to GitHub. The community is here so you’ll get more contributions right away in the form of ideas and feedback.

We do have biweekly dev updates which are somewhat related. But discussing individual issues here would likely duplicate discussions from Github issues.

Yes Bevy has a big advantage because its a library, so everyone who uses it is a developer and is able to fix minor issues. In case of Lemmy only a small fraction of users are developers, and even less know Rust. I watched the video, but its not easy to take concrete suggestions and apply them to Lemmy. Maybe the community reviews for PRs would be a good idea to get people familiar with the codebase. It also mentions using a project board so we should consider that. Though Im not sure how to select issues for each milestone because again, everything is up for grabs.

Here is an alternative Piped link(s):

https://piped.video/u3PJaiSpbmc?si=AtdmdalfFidWOJYq

Piped is a privacy-respecting open-source alternative frontend to YouTube.

I’m open-source; check me out at GitHub.

Well, for one thing federated message boards are incredibly niche to start with, and the pool of people willing to work on one for free in their spare time is bound to be tiny aside from language concerns. I know we all want the fediverse to be the hot thing that everyone uses, but that ain’t reality.

I’m not exactly seeing a massive contributor pool for sublinks here either.

At the scale of having a handful of contributors, it’s more likely random variance than due to the language you’ve chosen. The sample size is simply too small.

I can speak for myself - I know Rust very well but I simply don’t have the time to contribute to Lemmy’s code (I’m also spending some time already being an admin for Feddit.dk and I feel that is all I can muster).

Getting contributors to open source projects is never easy, regardless of programming language.

But I mean… there is no problem with competition right. Maybe it’ll turn out that sublinks will have more development with more contributors. But it has yet to be shown, there are also only 1 or 2 developers working on sublinks at the moment, if you check the github contributor stats.

Because people only contribute to projects they care about, can get clout from, or use and want to fix something that’s bugging them. Lemmy is just too niche right now.

If in five years there’s still just the two devs then you may gloat and tell me I was wrong and I will accept the shame

Yep. The job market isn’t as strong for rust, which is what that chart is showing you. Corporate acceptance != popularity.

Rust is #6 on the Stack Overflow developer survey in popularity. https://survey.stackoverflow.co/2023

Again, I don’t know why the community is insisting on making this a dick measuring contest for languages. People love rust. People love Java. I know people who still love Perl. I know one guy who really seems to look fondly on his Fortran days. they’re all fine.

Just be happy that someone is excited enough to write some code to make the fediverse a little more diverse and maybe cool.

any software is better with more reliability. Static analysis becomes more and more important as the code base grows.

You’re saying static analysis and reliability like they’re things Rust objectively has that Java doesn’t.

The only things Rust has at a fundamental level are a borrow checker and value types.

Value types can be nice for performance in certain situations (which aren’t nearly as important for a web server as fibers/coroutines). The borrow checker is nice for threading (which isn’t an issue in a restful web server design as previously stated). That’s about it.

(Safe) Rust and Java are both memory safe. I don’t think any is better than the other in this aspect.

It’s not possible to write (memory) unsafe Java code. All Java code is memory safe. Most rust code is memory safe.

Of course that’s not entirely true because of the possibility for bugs in the Rust compiler or the JVM.

Java and Rust both can call out to C libraries as well which of course… You get what you get.

Python for instance. At least Java has static typing

Python can have static typing these days as well. It’s just optional and controlled by the project.

You can write functions… A static function is not particularly different from a free function written in a namespace. You can even use a static import to allow yourself to call it without qualification.

I’m referring to functional programming concepts like anonymous functions and functions as parameters. Java finally added support for this with Lambda expression in Java 8, but because it took them so long, most libraries and frameworks that wanted to support backwards compatibility didn’t use them and still forced everything into classes and objects, meaning that most Java code bases you find are purely object oriented.

______ProviderBuilderProvider

idea what this is in reference to. I have a feeling your issue is more with the frameworks you’ve worked with than Java itself.

It’s in reference to the fact that unlike basically every other modern programming language, Java doesn’t let you do stuff like have default parameters for functions, instead forcing you to overload them:

In Javascript / C# you just have:

function DoSomething(int  howManyTimes = 10){
    //something
}

In Java you have:

function DoSomething() {
    int howManyTimes = 10;
    //something
}
function DoSomething(int howManyTimes) {
    //something
}

Which then leads you to recommendations of using the Builder Pattern, and now you’ve got the function / class you care about rewritten three times, plus a whole a builder class and a bunch of building specific functions, all of which are basically just a whole bunch of unnecessary verbose bullshit to approximate default parameters that are expressed more clearly with 2 additional characters in other languages.

I do think the developer experience with Rust is better in general though, so even if Rust was used elsewhere, it would still be liked more than Java, I would guess.

But we won’t know who’s right for another 10 years maybe :P

Are they forcing you to use their software?

If you just never make any mistakes you’ll live a mistake free life.

“Having a car with no seatbelt is not a problem if you’re a good driver”

LMAO

if you develop your java application correct, then you have no issues of any kind.

No issues of any kind? That’s absurd. The idea that you can “just” write “correct” code in any language and then you won’t have “any issues of any kind” is not any developer’s experience ever. Humans are just not reliable enough for that.

I mean, the same is true for Lemmy.

When it comes to servers, memory is often the pricy point. And Java is a lot more memory-hungry than Rust.