Remix vs Tanstack start
I've been using Remix for a while and really appreciate its loader pattern, but I've found actions/fetchers somewhat awkward to work with. I tend to prefer using React Query for updates.
I'm interested in learning more about the differences between TanStack Start and Remix/React Router. Specifically:
1. How does the data fetching model in TanStack Start compare to Remix loaders?
2. Does TanStack Start provide a better experience for mutations compared to Remix actions?
3. What's the most common way to deploy TanStack Start applications? Most of my infrastructure is on AWS (using SST with Lambda, S3, API Gateway, and Supabase).
Thanks in advance - TanStack Start looks like a great library, and I'm excited to learn more about it!
3 Replies
helpful-purple•5mo ago
Hey there!
The main difference in loaders, is that TanStack loaders are isomorphic (they run on server during SSR and then run on client during subsequent navigations). In remix, they only run on server
TanStack does not have an equivalent to Remix Actions directly, but if you are using React Query, it does integrates nicely with TanStack Server Functions
You can deploy it on Node (AWS) without problems. I wouldn’t recommend Cloudflare yet since there are some rough edges, but besides that, the other options are fine. (I tried CF workers and Vercel myself)
metropolitan-bronzeOP•5mo ago
Great, I actually think its a good thing that theres no equivalent to actions!
The loaders that run on the client are they proxied to the backend or do they actually run fully client side?
Thanks!
helpful-purple•5mo ago
One of the things I liked the most about TanStack (coming from Remix) was replacing Remix Actions with Server Functions + React Query. So I think I agree.
The loader itself runs fully client side, but if you put server functions inside (or other rpc functions), they will run on the server (like a fetch call)