comments on draft-ietf-idnabis-bidi

Harald Alvestrand harald at alvestrand.no
Thu Aug 6 22:46:35 CEST 2009


Alireza Saleh wrote:
> Harald Tveit Alvestrand wrote:
>> When integrating comments into text, some additional notes....
>>
>>
>>  
>>>> 12) The next sentence says: "In a domain name consisting of only 
>>>> LDH-labels and labels that pass the test, the requirements of 
>>>> Section 3 are satisfied as long as a label that starts with an 
>>>> ASCII digit does not come after a right-to-left label that ends in 
>>>> a digit."
>>>> This is not true.  See example b above.
>>>>           
>>> You are right. This needs to be documented; I did not test this case.
>>>       
>> I changed the sentence to say "the requirements of Section
>>
>> 3 are satisfied as long as a label that starts with an ASCII digit 
>> does not come after a right-to-left label" - I think this is true for 
>> all cases.
>>   
> I think we came to the consensus during the Dublin meeting that 
> IDNABIS working group should stay within the scope of labels and it 
> shouldn't do the intra-label checking.Bringing the new concept ( intra 
> label checking )  for domains will put registries into trouble.
Do you mean intra (within a label) or inter (between labels) checking?
> If the working group decides to do the intra-label checking then we 
> can think about some more complex rules such as : if a domain starts 
> with AL, then some of the current restrictions can be ignored however, 
> I think we can change the paragraph  which discusses about prohibition 
> of mixing the AN and EN characters in a label to something like  " AN 
> and EN cannot be mixed within a label if the labels doesn't start with 
> AL " .
Assuming you mean inter-label: This paragraph does not describe tests. 
It describes the properties of various domain names where some 
components pass the test.

We still don't require inter-label checks anywhere.
>
> I would also recommend the removal of the text from bidi that talks 
> about label uniqueness, the text itself reduces the accuracy of the 
> protocol and it may cause some inconsistency, as Harald mentioned 
> during the IETF meeting , they have a script that tests all possible 
> cases and those problematic cases have been already resolved by some 
> accurate rules, so if in the future we understand that something is 
> missing, then an explicit rule regarding them should be  added to the 
> protocol. The label uniqueness  can also be moved to rationale 
> document to cover all cases including homographs.
The label uniqueness requirement is, to my mind, necessary. It is part 
of the reason for the design of the test. Eliminating it while keeping 
the same rules will not do any good - it will make it more difficult to 
verify if any later change continues to satisfy the requirements that 
the original rules were designed for.

                    Harald



More information about the Idna-update mailing list