The [[syncing]] design assumes the network is connected. But it's often not in these pre-IPV6 days, so the cloud needs to be used to bridge between LANS. ## more cloud providers Git-annex already supports storing large files in several cloud providers via [[special_remotes]]. More should be added, such as: * Google drive (attractive because it's free, only 5 gb tho) * OpenStack Swift (teh future) * Box.com (it's free, and current method is hard to set up and a sorta shakey; a better method would be to use its API) * Dropbox? That would be ironic.. Via its API, presumably. * [[Amazon Glacier|todo/special_remote_for_amazon_glacier]] * [nimbus.io](https://nimbus.io/) Fairly low prices ($0.06/GB); REST API; free software See poll at [[polls/prioritizing_special_remotes]]. ## The cloud notification problem Alice and Bob have repos, and there is a cloud remote they both share. Alice adds a file; the assistant transfers it to the cloud remote. How does Bob find out about it? There are two parts to this problem. Bob needs to find out that there's been a change to Alice's git repo. Then he needs to pull from Alice's git repo, or some other repo in the cloud she pushed to. Once both steps are done, the assistant will transfer the file from the cloud to Bob. * dvcs-autosync uses jabber; all repos need to have the same jabber account configured, and send self-messages. An alternative would be to have different accounts that join a channel or message each other. Still needs account configuration. * irc could be used. With a default irc network, and an agreed-upon channel, no configuration should be needed. IRC might be harder to get through some firewalls, and is prone to netsplits, etc. IRC networks have reasons to be wary of bots using them. Only basic notifications could be done over irc, as it has little security. * When there's a ssh server involved, code could be run on it to notify logged-in clients. But this is not a general solution to this problem. * pubsubhubbub does not seem like an option; its hubs want to pull down a feed over http. ### jabber TODO * test with big servers, eg google chat * Prevent idle disconnection. Probably means sending or receiving pings, but would prefer to avoid eg pinging every 60 seconds as some clients do. * Make the git-annex clients invisible, so a user can use their regular account without always seeming to be present when git-annex is logged in. See ### jabber security Any data git-annex sends over this XMPP will be visible to the XMPP account's buddies, to the XMPP server, and quite likely to other interested parties. So it's important to consider the security exposure of using it. If git-annex sends only a single bit notification, this lets attackers know when the user is active and changing files. Although the assistant's other syncing activities can somewhat mask this. As soon as git-annex does anything unlike any other client, an attacker can see how many clients are connected for a user, and fingerprint the ones running git-annex, and determine how many clients are running git-annex. If git-annex sent the UUID of the remote it pushed to, this would let attackers determine how many different remotes are being used, and map some of the connections between clients and remotes. ## storing git repos in the cloud Of course, one option is to just use github etc to store the git repo. Two things can store git repos in Amazon S3: * * Another option is to not store the git repo in the cloud, but push/pull peer-to-peer. When peers cannot directly talk to one-another, this could be bounced through something like XMPP.