At 6:47 PM -0500 10/5/01, Eric A. Hall wrote: >My statement is supported. What I am seeing is policy direction such as >RFC2130, the protocols and formatting specifications under development >which are implementing UTF-8 as the preferred or only encoding, and the >number of platforms which are implementing UTF-8.
Every specification with a "preferred" encoding has one or more non-preferred encodings. Saying that UTF-8 is preferred is correct and irrelevant. As long as there are many popular protocols with lots of encoding choices, picking one of those choices for the wire format doesn't help any protocol except those that use that one encoding. Look at the popular protocols: they are almost all multi-encoding. Picking UTF-8 won't help any of them. Further, remember that UDNS requires that the UTF-8 sent to the DNS be marked specially. The cost of adding code for a new encoding for DNS (ACE) and adding code for a new way of doing DNS (UDNS) is similar. The coder needs to think about one or the other, and both will be a small chunk of code. > Items 2 and 3 in >particular illustrate that UTF-8 is and will be (for the "foreseeable >future" anyway) the favored encoding for new Internet protocols (and even >for new implementations, to the extent where other charsets are only >provided as subsets of 10646/Unicode). True, and irrelevant. >I'm certainly willing to change my position if you can provide evidence >that all of the above is just a bad expirement, and that NEW protocols are >in fact favoring some other encoding instead of UTF-8. Then you won't change your position, even when faced with other better arguments about why UDNS won't work. Let's try those again: - UDNS causes some strings that are legal in one encoding to be illegal in the other, and vice versa, meaning that some host names will be illegal part of the time, unpredictably - UDNS requires more work to be done by authoritative DNS servers - UDNS UTF-8 queries that fail will cause more load on the DNS - UDNS is not compatible with DNS security - UDNS requires that applications have the logic to send a new DNS query format - Applications that implement the UTF-8 part of UDNS have to handle the inevitable errors that come from queries that bounce, have to recast those queries in ACE (assuming that the name is even legal in ACE, which some won't be), and then have to emit those queries again, causing more DNS traffic - The errors caused by UTF-8 queries are the same as other legitimate DNS errors, meaning that applications will have to have their own (probably-nonstandard) logic for differentiating between expected errors and real errors UDNS gives us all of these problems for the limited benefit that some applications don't have to implement one additional encoding for sending host names on the wire. Does that seem like a good balance? --Paul Hoffman, Director --Internet Mail Consortium
