Pooya Parsa

This is kinda expected current behavior. Previously this workaround worked well because dependency chunks were seperated but now they are not it is a problem. We likely have to removal...

Thanks for the PR dear @elliotjarnit, it looks promising but i have to do more research specially to make sure it will be compatible with different providers.

Thanks for the report. Feel free to open a PR for adding support. Only we have to make sure signal is lazy created. Check https://github.com/h3js/srvx/pull/76 as reference

Can you please add a regression test?

Thanks for nice explanations. While it all makes sense, i am thinking more about a more generic hook name. For now marking PR as draft.

I believe #973 should fix encoding by keeping registered routes in encoded format. (if not please ping to reopen/discuss)

Thanks for explaining your situation. Have you tried force resolving to [jiti.cjs](https://github.com/unjs/jiti/blob/main/lib/jiti.cjs)? If you do it, webpack can transform predictable commonjs requires it only cannot determine ESM lazy import.

I guess you might try a webpack resolve alias like `{ jiti: require.resolve("jiti") }` in webpack.config (when using `require.resolve` it will be resolve to `node_modules/jiti/lib/jiti.cjs`) Benefit is, even without lazy...

Related: I am planning to replace old [http-proxy](https://github.com/http-party/node-http-proxy) used in [dev server](https://github.com/unjs/nitro/blob/1b9f51b7955d2490b1fd8da73af40780f30a1235/src/dev/server.ts#L257) with a better implementation. I will try to reproduce the issue with old package anyway but at the...

Sorry for delay on this. I guess now that we have official [WebSocket](https://nitro.unjs.io/guide/websocket) support, it should be less priority but I will defneitly investigate how to make it an extendable...