On 21. 06. 23 21:46, Salz, Rich wrote:
I have made the other changes, just seeking clarification on one point.
6.4. Matching an IP Address Portion
This document does not specify how an SRV-ID reference identity can include
an IP address.
I think SRV-ID clearly says it's just DNS name, so the presented identifier
cannot match an IP address. I think this section should clearly say that IP
addresses cannot match SRV-ID.
I am confused, because I thought an IP address *was* a DNS name.
It is, but the implication works only in one direction.
Here's my reasoning:
- Text representation of an IP address is a syntactically valid text
representation of DNS name. (Again, hostnames are subset of all DNS
names, so it probably is not a valid hostname.)
- Consequently, a blind attempt to put text representation of an IP
address into DNS resolution library will likely work and lead to
unexpected results.
- In other words, problem arises when we confuse the two data types,
e.g. when a program does not have explicit type (IP vs. DNS name)
associated with a given input.
- https://www.rfc-editor.org/rfc/rfc4985#section-2 clearly says that
"Name" component of SRV-ID is
> The DNS domain name of the domain where the specified service is located.
This does not give any leeway. The data type for the Name component is
defined as DNS name, even for cases where the IA5String happens to be a
syntactically valid IP address. I.e. it should always be treated as DNS
name.
Does that make sense?
--
Petr Špaček
P.S. As a side note, string "2001:db8::1" is _also_ a syntactically
valid DNS _name_.
Watch your inputs - e.g. function call dns.name.from_text('2001:db8::1')
from dnspython library will happily create a DNS name object from this
string, and will send it to DNS for resolution if you instruct it to do
so ...
_______________________________________________
Uta mailing list
Uta@ietf.org
https://www.ietf.org/mailman/listinfo/uta
