@TheLarkInn@djsmith42 I don't have any rigorous stats yet. These are only first dry runs on some of our small apps (<1mb ). Many of our large apps (>10mb 1000s of file) won't build with parcel yet. In a couple of weeks, after my assessment, I'll have detailed stats comparing them.
@TheLarkInn@djsmith42 I went back and checked these numbers are wrong. A different bundle was 74 seconds. Wepback was 6s and parcel was 3s. Thanks for keeping me honest.
@djsmith42 4) Some of its dependencies (e.g. uglify) are not mature enough to handle its vision. 5) There is a 50% it will become the standard bundler.
I am excited about its future.
@djsmith42 3) It will transpile node_modules by default respecting their .babelrc files. modules that are not prepared for this may cause build errrors. I hope this will solve the differences between the react-native and react communities.
@djsmith42 I have only experimented so far. My thoughts: 1) Webpack was a necessary exploration platform to figure out how to load/bundle non-js files. 2) Parcel is fast. I have seen bundle times from from 74 sec to 3 sec on initial builds. <1s with wet cache.
@djsmith42 I believe that the the current state of tree shaking is shaking is at the import/export level, but many framework APIs are at other levels like class methods. I do not believe those can be shaken.