yeoman
Discussion: The Future of `generator-webapp`
As part of the ongoing Yeoman Maintenance Reboot Initiative, we’re reviewing the long-term direction for generator-webapp — one of Yeoman’s classic generators.
This project has served developers for years, helping scaffold modern front-end web apps with build tools and optimizations.
However, since the stalled v4 planning (https://github.com/yeoman/generator-webapp/issues/759), the web ecosystem has changed a lot — and it’s time to decide what makes sense next.
💬 What We’d Like to Discuss
We’d love input from the community on the future of generator-webapp.
Should we:
Option 1: Deprecate
Formally deprecate the package and encourage developers to migrate to more modern, framework-specific tooling (Vite, CRA alternatives, static site starters, etc.).
This would mean no new features or maintenance, and the repository could eventually be archived and the npm package deprecated.
Option 2: Rebuild for 2025
Archive this repo (similar to generator-webapp_DEPRECATED) and start a new project from scratch, focused on:
- Classic, framework-agnostic HTML/CSS/JS
- Optional utilities like TypeScript, Sass, and testing setups
- E2E testing support
- Simple “old-school” scaffolding with a modern toolchain
This could keep Yeoman’s original purpose alive — generating clean, flexible web app starters without framework lock-in.
🧭 Context
- The v4 effort has been on hold for few years.
- Dependencies and build tools are outdated.
- The web ecosystem has moved significantly (bundlers, frameworks, etc.).
- It might be time to rethink the purpose of a “Yeoman web app generator” in 2025.
💡 How You Can Help
We’d love to hear from you:
- Do you still use
generator-webapp? - Would you prefer deprecation or a fresh reboot?
- What kind of “modern classic” stack would you find useful today?
All feedback — big or small — is welcome! 🎩
These generators served me big time when Node.js tooling for frontend started to bloom and was still difficult to set up. With Vite this is no longer the case, I would now like to defer to their powerful ecosystem of being able to orchestrate just about everything in a very clear way.
Recently I migrated a big project from a bunch of convoluted gulp tasks over to Vite. It felt right, and ended up being much faster, I got some things for free like HMR etc. And now my build step is so much easier to reason with and maintain So in this generator I really do not want to double down on gulp as our technology, IMO the tasks are not easy to follow, the ecosystem really needed a better build tool.
But I would not like to prescribe how to do a combination of things with Vite, and there are so many new things to support now… I sincerely believe that it's easy enough to figure this out today, perhaps with the help of AI. But before all we had were open source templates and perhaps some blog posts to use as references when trying to set up something.
With gulp the building blocks affect each other a lot, it's not easy to set them up. Also, at the beginning of generator-webapp technologies were much fewer, Sass was the de facto standard, Tailwind didn't exist, but now even in this one single area (styling) there are a few different technologies to choose from… let alone frameworks, and choices for those frameworks, and maintaining all of those, like Tailwind v4 and React Compiler… I feel that it would be too easy for the generator to become outdated. Or perhaps I'm pessimistic under the weight of JS fatigue.
On the other hand, today we have much better testing tools! If we were to improve generator-webapp, we could write tests that ensure that each combination of choices does what it should be doing, instead of… hoping that it does. The question just becomes how many choices there would be.
