Stephen Hahn wrote: >* Darren Reed <Darren.Reed at Sun.COM> [2008-07-30 19:45]: > >>Stephen, >> >>I've updated the issues files with the questions from the meeting >>this morning and made a brief note of the answers that I could >>remember. >> >>There were two questions I had from the discussion that I didn't >>bring up at the time: >> >>djr-3 Can package authorities be discovered rather than configured? >> >>djr-5 If multiple catalogues/depots are available, how does IPS choose >> which one to use if they are publishing conflicting information? >> >>For djr-3, I'm thinking along the lines of using multicast discovery on >>your local LAN or corporate WAN/LAN or maybe clues via DHCP or >>even a special DHCP tag or ... >> > > Yes, we think multicast discovery is very interesting for discovering > local depots. We'd also like to have a means for one repository to > offer pointers to other interesting repositories, although this could > be as simple as a package with a bunch of authority definitions. > > We'll discuss djr-5 and get a proper response, but fully adversarial > repositories, presumably with legitimate cryptographic tokens, hasn't > been a focus. Our model has been trust signed metadata, distrust > contents. We could go further into what "trust" means, I suppose. >
Thanks for taking these up, I'll look forward to seeing what you guys come up with. The main goal of these two is if I have my laptop that moves between home and Sun, it is highly likely that in the future there will be a depot on SWAN and highly unlikely I will have my own at home. My thoughts are that ips should be able to use the "closest" or "best" depot without me having to tell it every time I plugin. If this can be automated, it is important to have some analysis of the various threat models, from the adversarial to simply "old data" depot, that arise. Some of this may have ties with NWAM but I'm reluctant to suggest that this project should be dependant on NWAM. Cheers, Darren
