Tech stack for new frontend

[flosse]

Problem

Currently we have a mixture of a server side rendered frontend based on maud and a one page application based on seed. We should unify this somehow.

Opportunities

Here are some approaches I can think of:

• pure SSR (with maud)
  • ✔️ fast & reliable
  • ✔️ easy to maintain
  • ❌ UI too much tied to the server implementation
  • ❌ interactive parts needs to be injected with JS
  • ❌ extra routes to modify data
• pure OPA (with seed)
  • ✔️ TEA
  • ❌ seed is not maintained anymore
  • ❔ functionalities needs to be exposed in the public API
• use a new Rust frontend framework with SSR support like
  • dioxus
    • ✔️ popular
    • ❓ unclear future / roadmap
    • ❌ React style
    • ❌ No clear focus (too many platforms/targets/batteries included)
  • leptos
    • ✔️ fast
    • ✔️ popular
    • ❓ a lot of breaking changes
  • sycamore
    • ✔️ fast
    • ❓ overtaken by leptos
  • yew
    • ✔️ popular / maintained
    • ❌ slow
    • ❌ React style
  • sauron
    • ✔️ TEA
    • ❌ not popular / slow development
  • mogwai
    • ❌ not popular / slow development
• use a JS/TS frontend
  • ❌ JS/TS
  • ❌ npm and its ecosystem
  • ❌ hard to maintain

Conclusion

Currently (January 2023) it looks like leptos seems to be an interesting option.
There are risks (very young, few docs, breaking changes) but also great chances (suuper fast, lightweight, focused, fast growing user base).