12 replies

1. Tom MacWright 2 min later

   @zachleat a good starting point: https://github.com/wooorm/npm-esm-vs-cjs

   1. Zach Leatherman 3 min later

      @tmcw jinx! https://fediverse.zachleat.com/@zachleat/111812460993209987 (thanks!)

   2. Jared White (ResistanceNet ✊) 11 min later

      @tmcw @zachleat Those stats…aren't great. 😕

      (I'm of the opinion ESM Eveywhere All at Once can't come soon enough…)

      1. Dan Burzo 40 min later

         @jaredwhite @tmcw @zachleat I can imagine swathes among that “high-impact packages” dataset will not derive any benefit from switching over. As long as *I* can write and publish ESM…

2. Zach Leatherman 2 min later

   oooh https://github.com/wooorm/npm-esm-vs-cjs from @wooorm looks to be an attempt at answering this!

   1. Francis 🏴‍☠️ Gulotta 19 min later

      @zachleat @wooorm That's a lot more than I figured!

      1. Zach Leatherman 45 min later

         @reconbot @wooorm yeah, same!

3. Axel Rauschmayer 9 min later

   @zachleat Not that I can recall. But there is this: https://mastodon.social/@wooorm/111454830247362464

   1. Zach Leatherman 19 min later

      @rauschma yeah, very useful! Thanks!

4. Lea Rosema 11 min later

   @zachleat I'd also be interested in whether UMD is still a thing? Strictly, you wouldn't need it anymore. I appreciated that kind of module in the pre-ESM era, as it worked in CommonJS as well as in the browser via script tag and even in the legacy requirejs AMD module system. (I've never used AMD though)

   1. Zach Leatherman 51 min later

      @lea I need fewer module systems in my life 😅

      1. Lea Rosema 53 min later

         @zachleat same, it's really confusing in the JS ecosystem. 🙈