|
|
|
|
|
|
by segphault
1216 days ago
|
|
|
I am deeply sympathetic to Alex Russell's position on frameworks, but the reality is that he's never going to move the needle by badgering developers to prioritize performance over developer experience. The ecosystem won't move on from React to more performant and standards-based alternatives until those alternatives provide a competitive developer experience. As someone who wants to advocate for standards instead of frameworks in my org, the problem I have is that the standards process is moving too slowly to address very basic and obvious developer needs. For example, the fact that Web Components are registered in a global namespace is really inconvenient in a large organization where you have many different teams making a large number of components. Sure, you can work around this with a BEM-like naming scheme, but you shouldn't have to. There's a great proposal for scoped custom element registries by one of the Lit developers that would actually address this issue, but it's been sitting in a repo for years without any meaningful activity and there doesn't appear to be any momentum around implementing it. Meanwhile, this is a thoroughly solved problem in the React and Vue ecosystems. That's just one small example, there are a ton of other papercuts and annoyances that there's no hope of fixing in the foreseeable future if solving them is going to involve a years-long slog through the standards process. The DOM part spec that emerged from Apple's template instantiation proposal has the potential to provide an extremely efficient standards-based target for React-like frameworks that would solve a lot of the performance problems, especially if it's used judiciously for just the dynamic parts of the page in conjunction with server rendering and declarative shadow DOM. But nobody seems to be working on making that a reality, activity on that spec is pretty much dormant. React is going to keep winning until the standards furnish comparable ergonomics. |
|
|
Meanwhile frameworks and libs (even those who originally were very supportive of the idea of web components) have moved past WCs and are exploring approaches that WCs will never ever be able to provide such as granular reactivity, seamless frontend-backend integration, or even eschewing components as a rendering primitive entirely.
> React is going to keep winning until the standards furnish comparable ergonomics.
In all honesty, Scoped CSS + Nested CSS proposals coupled with https://open-ui.org would've solved 99% of what web components are trying to be.
Meanwhile we're about 20 years away from WCs as a finished thing: https://w3c.github.io/webcomponents-cg/2022.html