That’s why we don’t want to implement it in stdlib we want to define how a provider of time zone info should look like, so third party libraries can provide it, while the stdlib can offer functions that use it (and would error if no provider is specified). So you could have a nerves-specific provider that does something special.
Similar here. The stdlib doesn’t care how the provider gets the info - it might sync it in the cluster, sync it with some external resources (like current tzdata does) or compile it into modules (like old tzdata does) or do something I cant even think of right now.
As stated in the original post, this is not the thing we should be discussing here. How the provider will obtain the time zone database is beyond the scope of this discussion. We want only to explore what functions would require the time zone information and what would be the best interface for a library to expose it to stdlib.
