No-build: pipe dream, or near future?
summary
Bundlers are everywhere yet many dislike them—about half on Mastodon and a third on Twitter say they wish they didn’t need bundlers. Reasons for bundlers cluster into three concrete needs: transpilation (TypeScript still needs a stalled TC39 Stage 1 proposal despite Node running .ts), performance (tree‑shaking and minification bundlers still outperform browsers), and dependency resolution. Practical alternatives include import maps, JSPM and a custom unbundler that vendors deps locally via the npm dependencies hook—it works but still has CDN, asset and Node‑API compatibility issues.
AI-written summary · about
chapters (17)
- 0:00Welcome and bundler dilemma
- 1:33Bundlers frustrate developers
- 3:42Bundler fragmentation and leaks
- 5:51Reducing bundler complexity
- 7:36Why we use bundlers
- 9:00TypeScript native support
- 11:12CSS features and bundlers
- 13:39Performance narrowing gaps
- 15:26Dependency resolution gap
- 17:36Usability cliff of dependencies
- 21:10Failed promise of vendoring
- 23:16CDNs as dependency risk
- 25:45Import maps solution
- 28:59Unbundler approach
- 30:46Import maps limits and future
- 31:30Agents and web guidance
- 32:54Summary and call to action
