"Steve Snyder" swsnyder@snydernet.net wrote Fri, 31 Aug 2012 14:38:46 -0400 (EDT):

| I'm wondering about the benefit of running abridge on an IPv6 address. | | Since the big announcement last December that v0.2.3.9 supports IPv6 addresses for bridges, I've read a few comments to the affect that BridgDB doesn't understand IPv6 addresses. | | So... what is the state of publishing IPv6 bridge addresses? Will clients requesting a bridge address ever be given an IPv6 address? Can a client specifically request an IPv6 address (or IPv4 address, for that matter)?

There are two parts to this. One, the bridge authority has to handle IPv6 OR ports correctly, for example test them for reachability. Two, the bridge authority needs to communicate the bridge addresses to BridgeDB which has to do something clever with them.

What's lacking at the moment is deployment of the first part, i.e. upgrading the bridge authority to a (not yet released) version of Tor that can handle IPv6 OR ports in server descriptors. This will happen during the next couple of weeks. The benefit of running a bridge on an IPv6 address will then rise significantly. (The benefit of it up until now has been helping out with testing of the code and being able to hand out an IPv6 bridge address to users who need it.)

Regarding clients requesting bridge addresses, that's not exactly how it (is supposed to) work(s)*. The human being running the client gets hold of bridge addresses somehow and types them into a Vidalia window or her Tor configuration file. The way she does this vary. The bridges page [1] lists both IPv4 and IPv6 addresses to choose from.

(*) A client _can_ actually get an OR port from the bridge authority by asking for it using a hash of its identity (or was it descriptor?) but we're moving away from that IIUC.

[1] https://bridges.torproject.org/