Fork me on GitHub
#clojurescript
<
2022-07-15
>
Marcus17:07:03

Which web framework would you use if you were to start a new project today (something stable for the long term)?

danieroux18:07:09

Depends on what you care about. We did #helix for a long while, because it's a thin and clean wrapper on top of vanilla React. Clear path into that environment. Now we are on #fulcro and it is worth all the cognitive uphill we had to go through. It's all simple and smooth now. It holds complexity well, and the code is well written.

👍 1
Marcus12:07:37

Thanks! 🙂 I will give Fulcro a go.

👌 1
kennytilton15:07:00

Please ask again on the #clojuredart channel. Hint. 🙂 (I have heard good things about #fulcro as well.)

Marcus18:07:32

@U0PUGPSFR I just started down the Fulcro-road. 🙂 If I can't make it work I will check out clojuredart. Thanks!

geraldodev14:07:58

helix at client side is awesome. i'm into applied-science/js-interop. I consider a must, because we interact with js a lot and this library makes it very pleasant. I didnt used but uix is a thing of beauty. I really dont like hiccup for generating react interfaces. I'm sticking with helix but source code, attention to detail, documentation. It's beautiful.

geraldodev15:07:00

I've started the fulcro road. Long before I've started the Om road.That road was upgraded by fulcro. Fulcro is awesome if you brain is very good at trees.. Mine isnt. It lost me when at server side I need to use pathom (which is awesome), but I'm getting data from sql and pathom is not adapted to it. My template project is fulcro template, that I stripped fulcro and used the other bits. Fulcro code is very high quality. But there has to be mid term. Nowadays I'm thinking if my data is sitting at a sql database, I probably dont want to build nested data to edit it. I know react are trees already, but we program a component at a time. I do not think a client side database of stale information is good anymore. The cost of buiding it its high: Give up the efficiency of sql queries that brings the data very fast. I'm not facebook, I want to program fast with good quality. But I'm not going to rebuild graphql with clojure and give up years of sql efficiency. Nowadays I'm leaning to flat data that is brought from the server by react-query. Im playing with use react-query to instrument my async interactions with the server. The nice thing is that it does not fetch the data. All it needs its a function: Im using that function with lambdaisland/fetch to bring me transit that is served by awesome rest with reiti/ring/swagger to the client. Let's see where this roads gets me. I think clojure is awesome for data manipulation. But if we make everything a tree and say ppl that is fine and they need to use EQL, we are loosing ppl. Because this world needs datomic at server side.