Well, in the current API, you can either come up with a unique name, or not have any client-name-specific state i.e. either have no persistent requests or store them all on the global queue. I don't see how any other solution would help - if you start persistent requests, exit, another client steals your name, you give way to that client, then you'll have lost them.
On Friday 30 November 2007 13:48, Juergen Urner wrote: > Matthew Toseland wrote: > > > > Collisions result from one of two scenarios: > > 1) The client doesn't have a unique name. This is a bug in the client. > > 2) Two copies of the client are run simultaneously. FCP is intended to be used > > by one computer, clients which care about this can further uniquify their > > names by e.g. adding the local username. > > > > First of all, sorry for taking so much of your time. I am aware that > this is a corner case. > > case1: > if a client can not come up with names that are guaranteed to be unique > within the client its a client bug. True. > > case2: > Correct me if I am wrong, but I'd assume case two applies on unrelated > clients > aswell, not only on multiple instances of the same client. If my assumption > is true, clients have to be prepaired to deal with the fact that a collision > might happen the very next moment. Either this or the node takes over > due to the fact that it is the only instance that can guarantee uniquness > in this case. > > > Sidenote: > I know how to come up with a name that is highly likely to be unique. I > implemented RFC 4122 in Python and am pretty much aware of the drill > and the limitations of of any approach in this direction. > > > Juergen > > > > > > > > > _______________________________________________ > Tech mailing list > Tech at freenetproject.org > http://emu.freenetproject.org/cgi-bin/mailman/listinfo/tech > > -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: not available URL: <https://emu.freenetproject.org/pipermail/tech/attachments/20071201/21971007/attachment.pgp>