Post

that's what I'm trying to figure out it's not clear that the article is still up-to-date

a million percent; indulging people's irrational aesthetic preference for ".js" was always a grievous error the sole purpose of a file extension for a textual format is to determine the parse goal, and Module is different from Script, and `.js` should have only ever meant Script ¯\_(ツ)_/¯

📌

it was @kurtextrem.de 🙏

Wouldn't this be impacted by the "type" declaration in package.json too? I'd expect files with a ".js" extension to be optimized appropriately if I set "type" to "module".

Oh wow. This is really unexpected to here that ESM in *.js will disable optimizations. I hope it is confirmed as being outdated.

Yes I did. After writing the module, we received feedback from the V8 team (including that the doc you linked is outdated). The problem is that it fails silently so it doesn't show up in a trace. Tomorrow I can send you all the info I have as I'm not on my laptop. Cc @kurtextrem.de

I _think_ @kevinfarrugia.bsky.social thought the impact was minimal (but if it's even slight I think we should reconsider our default format in @nuxt.com and possibly also in @vite.dev). the key thing I'd like to confirm is that it really is the case first ...

thanks for sharing! i'm going to make esm.sh to follow that!

That’s interesting. In some of my tests esmodules (served as .js) were surprisingly fast compared with a bundled app, even. I’ll have to test with .mjs now