Code relating to the toggle code currently lives in
mobile.languages.structured and we will move it to webpack. It is currently lazy loaded meaning we are in uncharted territory for the migration.
- a conversation is had about whether we want to continue to code split by feature. Alternative approaches to consider - critical and non-critical js; load all code on startup. The conversation does not need to block implementation. At minimum we need to ensure that we do not increase the critical JS size
- code for feature is loaded in webpack (added topic to super happy dev time)
- tests are ported to node-qunit
Minerva is retained in a mergeable state throughout the migration. No need to worry about cached html but an alias module (empty module with mobile.startup dependency) will be needed until references in Minerva have been updated. JR: Wasn't applicable here.
- Critical JS has not increased (22B increase is tolerable)
- The bytes size of the language module has not changed considerably (e.g. we haven't loaded mobile.startup code twice)
Sign off steps
- Progress is updated.
- Unless I'm mistaken the port to webpack has shaved 2kb off the language overlay (post-gzip) without any changes to functionality. It would be interesting to understand how this came to be and record it for prosperity!
- Review which features to port next and create cards. (T210209 is next.)