We used to do this. It made the bundle size HUGE for not much gain. It can probably be done efficiently with selective, dynamic imports.

A solution should use https://nextjs.org/docs/app/building-your-application/optimizing/bundle-analyzer to make sure it doesn't bloat the bundle.

How about just importing the files for a few popular languages to tackle this one? If so, which languages would you want to support?

If they’re dynamically imported, it shouldn’t matter how many we support. But maybe we can find the top 5 somewhere

I'm not sure how you'd decide when to "dynamically import" a language file. Selectively importing languages on build would keep the size down but you'd be limited those that you choose. If you mean importing them on the fly then that's a whole different ball game.

If you mean importing them on the fly then that's a whole different ball game.

That's the ball game I mean.

Unless we use dynamic imports, this isn't worth the bundle size bloat. Maybe we can dynamically import all languages. It's been a long time since I've looked at the code, but I think each language can be imported only when they're used.

Well there is such a thing as dynamic imports but that isn't what you refer to as dynamic imports. You can choose which languages you want ahead of time but not during runtime.

https://mieszkogulinski.github.io/next-js-conditional-dynamic-import/

That's cool but once you've imported how do you decide when to drop it or does it just "exist" for the duration of the page view?

You don't decide to drop it. The browser caches it or doesn't. The main thing is that it's only downloaded if it's used.