Re: [DNSOP] why did SRV care to avoid conflicts

But the real reason for using _blah is so you can split ownership of
tcp.com from _tcp.com which you couldn’t do if the well know label names
used the preferred name format (LDH) is it would be “tcp” rather than “_tcp”.

> On 27 Mar 2018, at 9:18 am, Mark Andrews <marka@isc.org> wrote:
> 
> Type comes AFTER name when looking up records.  Label name spaces are NOT type specific.  They are relative to the  RHS of the name they are part of.
> 
> There needs to be a registry of all label that can appear immediately to the left of a valid host name.
> For each of those labels there needs to registries of what can appear to the immediate left of it.
> For each of those labels there needs to registries of what can appear to the immediate left of it.
> For each of those labels there needs to registries of what can appear to the immediate left of it.
> For each of those labels there needs to registries of what can appear to the immediate left of it.
> For each of those labels there needs to registries of what can appear to the immediate left of it.
> For each of those labels there needs to registries of what can appear to the immediate left of it.
> 
> EACH of these registries are INDEPENDENT of type.
> 
> For SRV records only the first two levels matter.  For other types there are different numbers of registries involved.
> 
>> On 27 Mar 2018, at 9:07 am, Paul Vixie <paul@redbarn.org> wrote:
>> 
>> following klensin's observations that cross-rrtype conflicts cannot occur and that the only _necessary_ registry is of predefined labels with specific meanings _within each rrtype_, i've strugged to piece together my memory of why i wanted _ names for SRV at all.
>> 
>> it's because the meaning of RCODE=0 && ANCOUNT=00 was once different, or at least interpreted differently, by some stub resolvers, compared to RCODE=3. it had to do with temporary vs. permanent failures in e-mail, as i recall.
>> 
>> in any case today it's a non-issue. if i were specifying SRV today i would not bother with underbars, because TCP.REDBARN.ORG or UDP.REDBARN.ORG would not confuse any app that looks up AAAA (or A), since they would hear RCODE=0 && ANCOUNT=0, and fail hard.
>> 
>> -- 
>> P Vixie
>> 
>> _______________________________________________
>> DNSOP mailing list
>> DNSOP@ietf.org
>> https://www.ietf.org/mailman/listinfo/dnsop
> 
> -- 
> Mark Andrews, ISC
> 1 Seymour St., Dundas Valley, NSW 2117, Australia
> PHONE: +61 2 9871 4742              INTERNET: marka@isc.org
> 
> _______________________________________________
> DNSOP mailing list
> DNSOP@ietf.org
> https://www.ietf.org/mailman/listinfo/dnsop

-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742              INTERNET: marka@isc.org