As I understand DNS-SD, it has no concept of endpoints (as we have in the RD), but only of services.
Is, therefore, the creation of {ep}.{d} host names for IP addresses (Section "Domain mapping") anything more than one suggestion of how an RD-DNS-SD server can arrive at FQDNs to put in the SRV records (because the SRV can't have plain literals) if the mapped URI doesn't already have a host name in its authority component?
I think it is; then the respective section should say:
Instances need canonical host names to refer to in their SRV records.
If the instance's target URI does not have a FQDN in its authority component, the RD-DNS-SD server needs to register an appropriate DNS name for the authority address as an A or AAAA record.
One way to derive such names is to prefix the "ep" attribute of a registration to the domain name for all such endpoint registrations, and to assign the address given in its "con" attribute to that name.
This text also emphasizes that this interacts with data from the endpoint lookup.
As I understand DNS-SD, it has no concept of endpoints (as we have in the RD), but only of services.
Is, therefore, the creation of
{ep}.{d}host names for IP addresses (Section "Domain mapping") anything more than one suggestion of how an RD-DNS-SD server can arrive at FQDNs to put in the SRV records (because the SRV can't have plain literals) if the mapped URI doesn't already have a host name in its authority component?I think it is; then the respective section should say:
This text also emphasizes that this interacts with data from the endpoint lookup.