Closed dominikg closed 1 year ago
Dropping Node 14 makes sense. If we dropped Node 16 that would prevent kit
from upgrading: https://github.com/sveltejs/kit/blob/c4efbc7ad957465bfb14723cdb00a6923ed19e5b/packages/kit/package.json#L100
sveltekit users can use node 18+ with vite-plugin-svelte 3.0, it won't prevent kit from upgrading, just use ^2 || ^3.
but sveltekit has a self interest in releasing a major with node18+. we can remove the undici dependency and rely on it being available (all node 18 versions have it, if we start with the most recent one the worst bugs should be gone).
it won't prevent kit from upgrading, just use ^2 || ^3
I suppose that works. Users creating new projects will end up with 3, but I suppose it's not a breaking change in that case since they're new projects. I'm still a bit nervous people will upgrade their lockfiles and end up with an incompatible version for existing projects and won't understand how to deal with it
sveltekit has a self interest in releasing a major with node18+
Agreed! I can't wait. Though I'll be even happier when we can stop polyfilling crypto to get rid of all the polyfills, but I think that requires Node 20
I'm not against having a SvelteKit 2.0 that requires Vite 5, vps 3, and Node 18. It might be a nice way to make the transition to Svelte 5 more gradual by making sure everyone has upgraded vps to the latest version and that all their libraries have valid entry points. I think incremental upgrades are easier than bundling all changes together
Describe the problem
supporting eol versions of node prevents us from updating some dependencies. Node 16 is going to be EOL very soon and vite 5 is not going to support it anymore
Describe the proposed solution
align with vite supported node versions
Alternatives considered
none
Importance
would make my life easier