User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.4) Gecko/20030624 Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.4) Gecko/20030624 When users entered Alphabet or Numerics in compatible (i.e Full-Width) form, the resulting ACE label (both Punycode and RACE) includes 0x0 in it. For example: www.U+FF4A U+FF44 U+FF4E U+FF41.jp is expected to be nameprepped and changed into www.jdna.jp, but the result is www.xn--jdna-.jp. Reproducible: Always Steps to Reproduce: 1. enter www.ｊｄｎａ.jp at address bar 2. 3. Actual Results: the entered address was changed into www.xn--jdna-.jp, and DNS fails Expected Results: ｊｄｎａ part should be nameprepped into jdna, i.e. www.jdna.jp
Confirmed. When we use Japanese Roman letters (either upper case or lowercase), NamePrep does not produce the correct result.
Another example is "maß", i.e. "bei U+00DF"; during NamePrep, the german double s is converted into two s and no non-LDH character remains. So it should not be converted to Punycode after the normalization, opposite to how it is done (xn--mass-).
Sorry, the example went wrong: it is maß -> U+006D U+0061 U+00DF
We branch into IDN code whenever a hostname contains non-ASCII codepoints. So, it seems to me that as a last step, our IDN converter needs to verify that the nameprepped hostname also contains non-ASCII values before we tack on the xn-- prefix. If I am correct, then this should be fairly easy to fix.
Darin, I tried your patch on Mozilla 1.4 branch with the example filed by the reporter. It works when the patch is applied. I also tried a case when the user enters www.(JPN name).jp (in JPN full- width Roman characters), this too works with the patch. Klaus, Yoneya-san, and others, please supply other real life examples for verification. Thanks.
Comment on attachment 127174 [details] [diff] [review] v1 patch r=nhotta
Comment on attachment 127174 [details] [diff] [review] v1 patch bz: please see comment #4 for an explanation of this patch.
I wonder if full-width alphabet only domain names are accepted for registration. I am not sure about the official guideline but I think that could be confusing (also a risk for spoofing). Yoneya san, does JPNIC allow those string for the registration?
This is re-comment to comment #10. JPRS (is JP registory since Apr 2002) accepts full-width Alpha-Numerics in inputted string for registration. But those are NAMEPREPped by registration system at the registration time and never appear in registered domain name. Allowed Japanese characters for JP domain name are listed on draft-yoneya-idn-jachar-00.txt
i think this patch should be included with any future 1.4 branch releases.
*** Bug 211320 has been marked as a duplicate of this bug. ***
Comment on attachment 127174 [details] [diff] [review] v1 patch this is an important bug to get fixed on the 1.4 branch. it prevents IDN from working with many sites.
Comment #15 is so true that we should definitely check this into 1.4 so that any future releases from 1.4 can benefit from this fix.
Thanks to all, I found this bug was fixed in 20030708 trunk. Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5a) Gecko/20030708
Comment on attachment 127174 [details] [diff] [review] v1 patch a=asa (on behalf of drivers) for checkin to the 1.4 branch.