[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: circuit recovery

i was aware of the 'long lived nodes' thing for persistent
connections... perhaps rating this on a longer timeline would be
helpful, an uptime average rather than current age - or have i misread
the spec in this regard?

i'm not quite sure i understand the thing of the circuit queues thing
you talk about in that faq entry - a 2/3rd improvement in connection
stability on irc would be very welcome in any case... i am speaking as a
user, and not as a developer, but i was under the impression relaying
did not have any process on the server side apart from listen and pass
forward... and thus the circuits built for connections to these long
lived connection types - perhaps they could be constructed beforehand in
such a way that they can be migrated periodically, and perhaps as a
further measure, if a more stable but less anonymous proxy node were
available, there could be some way to have the entire circuit shift,
since the endpoint proxy relay is almost guaranteed to be stable.

just putting in some thoughts, my primary usage of tor is for irc, so
this area is of much interest to me.

On Sun, 18 Dec 2005 01:11:40 -0500, "Roger Dingledine" <arma@xxxxxxx>
> On Sat, Dec 17, 2005 at 07:38:23PM -0800, Glymr Darkmoon wrote:
> > hey, i use irc a lot via tor, and one of the things which has been
> > bugging me endlessly is when circuits die
> > 
> > is there some reason why, unless exit nodes die, that a new circuit
> > ending in the same exit could not resume the same connection stream?
> > 
> > having persistent connections able to switch paths while still connected
> > would be immensely useful
> I've just fleshed out the FAQ entry on this:
> http://wiki.noreply.org/noreply/TheOnionRouter/TorFAQ#MigrateStreams
> Thanks for the nudge.
> --Roger
  Glymr Darkmoon

http://www.fastmail.fm - Or how I learned to stop worrying and
                          love email again