Dave Crocker wrote:
> >Also note that this decoding will require changes to some parts of > >the infrastructure in order for it to be a successful encoding, > >probably at the authoritative servers for a zone and possibly at the > >resolver depending on depth of adoption. > > The only changes need to be up the stack, towards user interactions. > That means the DNS client and the DNS Administration interfaces that > creates DNS entries. No other part of the system needs to change. This is patently false. Under the current spec, hostile redirects are trivially easy. To use your favorite phrase, the proposal is "incomplete" without guidance and at least a recommendation on these and other critical issues. At best, it will require application involvement, which is substantially more effort than simple display decoding. The depth to which application interfaces go in this and other efforts will also affect whether or not resolvers need to be changed. It was illustrated last week that answer data cannot be properly encoded in all cases. There are only a couple of ways to solve this problem, one of which is ensuring that all of the authoritative servers for a zone follow the same default recommendation. New servers. Quite simply, if IDNA only served the principle goal for which it was designed -- backwards compatibility with hostnames as transferred in application data -- then it would work without any other changes. Because it goes beyond this scope into display decodes and zone data, it is by definition no longer serving the single goal for which it was designed, and it inherits the problems associated with those turfs. Your failing to notice these problems doesn't mean they do not exist. You have made it clear that you despise basic science, but it should be equally clear that it is far too early to be annointing anything the winner in this process. Doing so can serve no purpose which is beneficial to this WG or the Internet community. -- Eric A. Hall http://www.ehsco.com/ Internet Core Protocols http://www.oreilly.com/catalog/coreprot/
