Evaluate a strategy to avoid losing user's language preference
We could build 12.0 both as a multi-lingual + "localized" to catch users that we won't catch with #41413 (closed):
- we build multi-lingual packages and installers: we'll upload only these ones to dist.tpo/link only these ones from www.tpo/download, so all new users are forced to be multi-lingual users
- we build localized packages only to make the incrementals:
- they will still contain all languages, so they are forced to be multi-lingual, too, and we move them to the
ALL
channel - we will add a special preference with the old channel, and we'll use it to make
intl.locale.requested
a user preference with the chosen language
- they will still contain all languages, so they are forced to be multi-lingual, too, and we move them to the
Pros:
- we might catch users that don't update every version, and keep their language preference even when switching to multi-lingual
Cons:
- time consuming to build, and possibly also to sign
- we'll have the same problem of incrementals we had with 12.0a3 -> 12.0a4 with 12.0 -> 12.0.1
- what about people that also miss 12.0? Should we continue doing these while we keep receiving a high number of update pings from old versions?
Edited by Pier Angelo Vendrame