Post

Who's got frontend frameworks? Trying to collect some stats on update efficiency (rather than rendering new dom en-masse (which js-framework-benchmark heavily favors <3))

I am actually trying to put my unpublished dumb little frontend framework through your benchmark if the results turn out below laughable, I might even publish this so we can all have a nice laugh at my expense 🤪

I've been building this little thing for 2.5 years mostly because I pitched the idea to Ryan (Solid) one time and he told me this wouldn't be ergonomic it's signals, but every dependency is declared, no tracking, no global state and it's fairly ergonomic 🤷‍♂️😄

I don't want to build or maintain a framework honestly I'm just trying to get it to a state where it gets the point across in part because I think the ECMA signals proposal is the wrong level of abstraction you can build tracked signals on untracked signals, but not the other way around 😌

I agree about tc39 signals being... too small a change proposal. The thing I'm worried about is: in order to have performant signals, you need batching and scheduling and like, signals without a way to render efficiently is just asking for performance problems.

0 reposts0 likes

since ember used to have manually tracked signals (they called them computeds, and were a huge source of bugs because people couldn't manage the explicit lists) -- I'm very against them myself lol. But!!!! I super appreciate the tenacity to make a whole project to prove a point <3 love it!

I only vaguely recall trying it many years ago, can't say how similar I guess we'll see, I just need to get it finished enough to show hopefully soon 😅

it's more the other way around -- efficient signals implementation must have batching and scheduling, else the fine-grainedness causes too many updates to "the view" faster than what can be perceived, and wastes CPU