@gweesin You probably already know this but I think @leo was previously merging on this repo but has left Vercel (if his twitter is any indication). Not so sure this is getting merged unless someone else from over there has the ability to approve. Maybe @rauchg or @TooTallNate knows who hits the buttons on this repo now?

I indeed don't have access to this repo anymore, but I just pressed "View git blame" on the line that imports fast-url-parser, which took me to this commit. That commit replaces the native url module with fast-url-parser.

I was wondering why (I don't remember the reason), but looking at the readme of the fast-url-parser package and considering it has 4 million weekly downloads, it seems that the only reason why the package exists is to make the native url module 25x faster (e.g. for parse()). That's why it has the same API too.

So unless the native url module got a lot faster over the years (which is probably the case, but we'd have to test to what extend that is the case), using it might make serve-handler a lot slower in production.

But agreed that, if the owner of fast-url-parser is not responsive, replacing it in server-handler might be the only option. I think it is very likely that the native url module has gotten faster over the years.

I've run fast-url-parser benchmarks (higher is better) with Node v22: fast-url-parser is 2.5 times faster than node:url.

For your information, here is the benchmark displayed in fast-url-parser's README dating back 11 years. fast-url-parser was 10 times faster.

That seems like a reasonable tradeoff to me, considering that using the native module would also make the package lighter.

Great to see that Node has gotten a lot faster. I'll try to ping someone at Vercel to merge this PR.

Thanks for the PR @gweesin and thanks for the benchmarks @regseb!

I'll try to cut a release later today with those changes

Thank you all for hopping on this. Really appreciate it.