McDonald, Ira writes: > It's likely that the attempt to retroactively introduce case sensitive > domain names will simply break almost all deployed Internet software.
You are misreading the proposal. Software will treat uppercase ASCII and lowercase ASCII as equivalent, for interoperability with existing configurations. However, software will not adopt the new insensitivity rules for non-ASCII characters. Uppercase Beta, for example, won't be automatically converted into lowercase beta. Registries will prohibit uppercase non-ASCII characters. Users will stick to lowercase characters. Consequently we'll have the flexibility to add case-insensitivity later _if_ that turns out to be a good idea. Case insensitivity is quite a bit of added work for implementors, and it dramatically expands the number of opportunities for visual confusion, as well as the severity of the confusion. The only counterargument is the unjustified speculation that users are too stupid to handle a world of case-sensitive names---even though URLs are already case-sensitive, specifically in path names. Do you think that the unjustified speculation outweighs the big problems? Are you so sure about this that you support making an irreversible decision right now? The careful approach is to start with lowercase IDNs. Registries will allow a carefully selected set of characters. It will be possible to expand the set later, when we have more experience, but for the moment we'll stick to something safe. Similar comments apply to Chinese. ---Dan
