Re: Send-N draft: draft-bellis-enum-send-n-02 published

On Jun 24, 2008, at 11:29 AM, Duane wrote:

> [email protected] wrote:
>
>> I've made specific allowance for an absolute form of the Send-N  
>> record
>> which *is* compatible with wildcard records.
>
> Yes, but it's still possible to break wild cards:
>
>> 7.2.  DNS Wildcards
>>
>>   The relative form of these records SHOULD NOT be used with DNS
>>   wildcards since DNS wildcards can represent an arbitrary number of
>>   labels (or digits, in the ENUM case) and the data in a relative  
>> form
>>   'Send-N' record is specific to an exact position in the ENUM tree.
>
> and if there is anything I've learnt, when ever you give someone  
> enough
> rope they'll surely hang themselves sooner rather than later.
>
> I fail to see why your use for DNS should take precedence over  
> existing
> DNS types at all, this will only lead to disparity between  
> countries, I
> find the idea of databasing this information useful, but don't break
> existing systems to achieve it.

I find the argument of 'X breaking wildcards' ridiculous. There is no  
such thing. If there would be anything breaking wildcards, it would  
have been used and deployed widely already, since wildcards are often  
considered evil.

Also, the concept of wildcards in DNS is often misunderstood. On many  
occasions, the DNSEXT WG were asked to please fix wildcards, as it did  
not behave as expected to those who wanted to use them.

Wildcards are a part of the dns protocol. NAPTR records are used by  
applications that use the DNS protocol. There is no way that  
application use of the DNS can break the underlying DNS protocol.

It is perfectly fine to state that X SHOULD NOT be used with  
wildcards. I would suggest you actually help to find alternative  
language to state that the use of the relative form of send-n NAPTR  
records combined with wildcard domain names is NOT RECOMMENDED as it  
might lead to unexpected behavior.

As for prior art, there are many other examples where use of X and  
wildcards are not recommended. DNAME and wildcard domain names. SRV  
records and wildcard domain names.

As for you 'precedence' statement, I fail to see how you come to the  
conclusion that relative send-n NAPTR RRSets have precedence over  
wildcards. Obviously, quite the reverse has been stated, in that these  
records SHOULD NOT be used with DNS wildcards. (i.e. if you have  
wildcards, you should not use these records). If you'd like to argue  
that this statement is too ambiguous, send Ray text to help him  
disambiguate the statement.

Furthermore, if there is any evidence that an application use of  
existing DNS resource record types breaks Wildcard Domain Names or the  
processing of it, or undermines the DNS protocol, please bring it to  
the DNS Extensions WG.

Thanks,

Roy