Roadmap: 2025-26

[benwis]

It's probably time for a new roadmap for the next year, I imagine Greg and I will fill this out in due time, but if anyone has ideas or requests they'd like to be considered that'd be suitable here, feel free to comment below

Build Tooling

• Rewrite cargo-leptos to be more ergonomic and easier to maintain
• Integrate wasm-split for webassembly splitting
• Explore dx subsecond integration

[christhirst]

I would love to have the projects updated and able to run. Especially the one with sessions. I tried the example with:
https://github.com/leptos-rs/leptos/tree/main/projects/sso_auth_axum
And I would prefer something like:
openidconnect-rs
Until now I was not able to get it running.

[gbj]

@christhirst The projects directory is explicitly not maintained, except to the extent that users want to contribute updates: it exists as a place for people to contribute things they've worked on without the expectation that I will keep the examples in there up to date over time. It should not be difficult to update any of them (relative to migrating between versions in general) but it is definitely not something that is going to be on the roadmap.

(See the projects README for the distinction between this and examples)

[dxps]

WASM splitting is a pretty appealing feature! 😍 At least it gives to some of us the confidence that we can build a website of any size and not get concerned about the generated WASM size. More to this, it could also provide support for microfrontends based solutions.

[medanimohamedakram]

would merging reactive_stores in the main leptos crate be possible

[dmgolembiowski]

@benwis I like the cargo-leptos idea. Here's my 2 cents.
At times the starter project templates had a tough time remaining consistent with upstream work in a compatible way. Any time a new project gets made with the latest improvements in mind, the boilerplate and dependency management gets tricky. And occasionally I found it difficult to use 3rd party things because of prioritizing feature completeness over upstream adherence.

IMO these can be really hard problems to solve, considering you and greg want to keep things as light as possible.

So here's my thought -- workflow actions that create tagged branches of the template projects that correspond with upstream improvements, then -- and here's the magic -- leverage a tool like degit or something with cargo-leptos to only copy source files that correspond to branch names like ... leptos-3/leptos-2/leptos-3-islands/leptos-3-ssr/whatever: so that building a new project doesn't have to be like cloning linux.

Personally, I would love to help implement this with you all the way.

[bicarlsen]

For me the biggest issue is compile and hot reload times for larger apps. I'm not sure where the slowness is coming from: rustc, trunk, tauri (I moslty build Tauri apps), or leptos. Perhaps this is something to look into and work with the other tools used by the Leptos community to see where to improve.

I tried the tricks of type erasure and splitting the app into different crates, which certainly helped, but it still takes 4 minutes for my app to recompile, even if I only change a single CSS class.

I'd be happy to contribute on this front, but would need some guidance.

[sebadob]

@bicarlsen Have you changed your linker already? If not, doing so will probably be the biggest improvement for you. If you are running linux, check out mold, and lld otherwise.

---

Regarding the Roadmap, I think the wasm splitting is probably one of the biggest improvements we can have. Really looking forward to it.

[bicarlsen]

@bicarlsen Have you changed your linker already? If not, doing so will probably be the biggest improvement for you. If you are running linux, check out mold, and lld otherwise.

Yes, I'm on Windows so tried with the LLD linker, but it still takes minutes to hot reload changes.

[NicTanghe]

Hi I think a larger library of oficial examples whould be a great adition for people new to the library.

I dont think using external ui libraries or widget libraries belong in these examples.

I had to rely on allot of help from all of you wich i am extremely grateful for. And i think haveing a list of wel written examples of just about everything and every combination of things. whould be of great help for people starting.

axums examples i found where the best written and usable examples that where easyest to replicate and grasp. And probably a great referance.

[rakshith-ravi]

If it's not too much to ask, a type-safe router would be super helpful! Something where I can mention the path parameters, the query params allowed and the ability to nest routes is very useful to make sure we redirect to the right URL.

Some inspiration:

• Dioxus does some good work on this front: https://dioxuslabs.com/learn/0.6/guide/routing#creating-a-route
• I personally LOVE yew's implementation where you can nest routers: https://yew.rs/docs/concepts/router#nested-router

Happy to help implement this if I can get some guidance on how to go about this

[maciek134]

I tried the tricks of type erasure and splitting the app into different crates, which certainly helped, but it still takes 4 minutes for my app to recompile, even if I only change a single CSS class.

I on the other hand would love it if build time optimizations stopped being the priority - that already killed generic renderers (your problem is not linking btw, -Z self-profile shows linking takes ~100ms on a hot build with a single changed class with mold).

Maybe some docs on how to structure the app to avoid the issues?

[bicarlsen]

(your problem is not linking btw, -Z self-profile shows linking takes ~100ms on a hot build with a single changed class with mold).

Indeed it was a configuration issue (see Issue #4185) and my app is compiling in seconds rather than minutes now. While compile times are no longer a main issue, I didn't retract the suggestion because others seem to agree.

[bicarlsen]

I'm also not sure if this is from how I have my app architected or something else particular to my setup, but I just saw @gbj's demo of hot reloading Leptos in a web app and would love that for Tauri apps, if it's possible. (I've only use Leptos for Tauri apps.)

Currently my Tauri app will fully rebuild anytime I change anything in the view macro.