Finished a port of a complex side project from #nextjs to #remix and I’m honestly quite happy.
Layouts are great, I had two failed ports to nextjs app directory. I’m not in love with remix’s route api but it works! I don’t have strange edge cases and there’s a js api escape hatch
loaders and actions are wonderful, I got rid of a whole ass api server. Having an actual request and response to work with is a breath of fresh air
next/image was fantastic I will miss it but also it wasn’t that hard to size my images
serverless to a server - this isn’t remix or nextjs related but this change drastically improved my websocket integration complexity and perf which is important to my application.
remix docs are smaller because it’s a much smaller api surface area afaict. This was great
I didn’t use vite and I should have, the move doesn’t look to painful and then I’d get a cleaner dev experience with more tooling. Haven’t invested in non product stuff yet.
Took me forever to get over compiling static html in favor of cache headers on dynamic responses. I have a logged in app so it was actually easier and cleaner to go this route and cut down on api calls. I didn’t need it.
Overall it’s been a great change and it’s working well for me and my users 🙌
I miss the centralization and middleware of the api with graphql/trpc but I’ve relegated a directory to be that layer and gave the mutations a common api structure that ensures a authz is at least front and center. I imagine this might mature to a wrapper function that requires it.
@janl It somehow feels not very "professional" but I don't know what else to do without some 3rd party service like turnstile, which I bet 90% of doesn't have to be a service at all.
Add comment