When we have an Internet connection we ideally want to support connecting to others and having them connect to us over Tor. This requires us to host a Tor Onion Proxy. This requires extending the replication manager so we can specify the URLs we want to push replications out to and for those to be tor aware URLs. This also requires creating our own hidden services so we can be reachable and deciding how often we will run those hidden services, especially when we are on battery constrained transports like cell. We also need a standard way (read: PouchDB) to communicate to others when we will be available (e.g. we promise to try once an hour for five minutes at 10 minutes past the hour) so they can wake up to try to push data to us. This also means replication managers need their own logic for when they will try to reach out to someone they need to push data to based on their declared availability.
Koffeegod
commented
8 years ago
When we have an Internet connection we ideally want to support connecting to others and having them connect to us over Tor. This requires us to host a Tor Onion Proxy. This requires extending the replication manager so we can specify the URLs we want to push replications out to and for those to be tor aware URLs. This also requires creating our own hidden services so we can be reachable and deciding how often we will run those hidden services, especially when we are on battery constrained transports like cell. We also need a standard way (read: PouchDB) to communicate to others when we will be available (e.g. we promise to try once an hour for five minutes at 10 minutes past the hour) so they can wake up to try to push data to us. This also means replication managers need their own logic for when they will try to reach out to someone they need to push data to based on their declared availability.
Make sure to review http://www.icsi.berkeley.edu/pubs/networking/asynchronousneighbordiscovery08.pdf, sent to us by Michael Rogers of Briar, to get ideas on how to best pick availability times.