On 11/12/10 1:12 AM, Stephane Bortzmeyer wrote: > On Thu, Dec 09, 2010 at 10:48:03AM +1100, > Ian G<i...@iang.org> wrote > a message of 39 lines which said: > >> I'd suggest there is a better path: >> >> Bypass the DNS. Work with IP#s directly. > > Remember that what DNS buys is *not* user-friendly names (even if they > are nice), it is stability. IP addresses change much more > frequently. Today, my blog is hosted at Slicehost, tomorrow, it will > be at OVH and the IP address will be different while > www.bortzmeyer.org will still work. > > Any solution which pretend to replace the DNS must address this > issue. Or state clearly that it implies PI addresses.
Of course. >> b. the client has to be resiliant to fall back through that list and >> try different IP#s. > > How the client will know it has reached the right one? Of course, if > one address yields "No route to host", you know it was not the right > one. But if it reached a target, how do you know it is the right one, > if the IP address was reallocated. High resiliance engineering (an assumption of the original post) typically involves private/public key pair authentication in both directions. Nothing happens if we're talking to the wrong party. >> PS: apps don't care about names or numbers, only humans get >> flustered about it. DNS is for humans, not apps. > > Completely wrong. If cron does a 'wget > http://code.google.com/p/libjingle/source/browse/trunk/talk/p2p/base/pseudotcp.h', > 'code.google.com' is not here for the pleasure of humans but to be > sure we go to the sam place, even if Google decides to change its > hosting system. I wouldn't have classed that line of code as anywhere near high resiliance engineering :) iang _______________________________________________ p2p-hackers mailing list p2p-hackers@lists.zooko.com http://lists.zooko.com/mailman/listinfo/p2p-hackers
