We aim to keep Meteor’s core features while always offering a modern setup path. We don’t want Meteor projects blocked by updates, nor do we want to lag behind JS standards, popular tools and the wider ecosystem.
That’s why our approach to updating Meteor’s bundler is modern-first. Since we’ve recently experimented with SWC integration, here’s how it works:
1. Measure goals
- Automate issue detection so we quickly see if we meet our standards later.
- In Meteor 3.2 we added
meteor profilefor build-time and bundle-size checks and also tried CPU profiling.
2. Try modern tools
- Dedicate time to integrate existing, popular solutions rather than rewrite everything ourselves.
- In Meteor 3.3 we began swapping Babel for SWC to remove our biggest build slowdowns.
3. Validate results
- Rerun our measurements to check against standards.
- In Meteor 3.3, early tests showed SWC cutting build times significantly, so the process could continue.
4. Handle regressions and compatibility
- Address any Meteor-specific limitations in the integrated tool and offer flexible workarounds when possible. If a workaround isn’t feasible, hide the feature behind a configuration option. Leave more complex fixes open for us or the community to tackle later.
- In Meteor 3.3 we found that top-level await and HMR were broken initially. We preserved reify for solving those, since that was an easy fix. We couldn’t keep nested imports, because these aren’t standards for modern syntax tools, so we need an SWC plugin to parse import statements inside blocks; don’t confuse nested imports with standard dynamic imports, which SWC still supports its syntax.
We also added configuration options to opt out of SWC transpilation and minification, and a fallback to Babel for backwards compatibility.
We stay in this process because re-implementing features modern tools already cover can stall our PR indefinitely and drain our motivation, community tools already solve these problems and maintain solutions over time.
Refer to the previous answer for the current status of the Moder Bundler solution integration.
Back to your question, we can’t yet specify exactly what will be preserved or discarded in a modern setup or package. We will continually develop opt-in/out options so each project can mix old and new as needed. The modern bundler will likely handle only your app code, with configurable exceptions, and Meteor-specific features will remain in Atmosphere packages. This still depends on what the experimentation phase shows us. If we can preserve features in modern setup, we will, just as we did with SWC, or we’ll find alternatives. Meteor 3.3 also focused on Meteor-specific optimizations. We could have started with a modern bundler integration, but we chose to improve existing projects and support those who prefer the Meteor bundler strictly. Modern bundler integration will also benefit from these optimizations.
We don’t plan to work on Reify unless we have time, know how, and once a modern component is integrated first. Standard JS features like dynamic imports will be supported, perhaps not exactly as Meteor does, but your project should notice no difference except that DDP won’t run behind the scenes. Preserving nested imports is harder, anyone can help add support in SWC. Meteor 3.3 detects nested imports automatically in your code using verbose mode so you can adopt standards easily if you prefer. We haven’t added a breaking change to your project but provided a Babel fallback so you can still benefit from improvements. This is our way forward with modern bundler integration as well.