Moin!

On 14 Nov 2026, at 23:56, Joe Abley wrote:
> Imagine a web server sends mappings between names and addresses to a client 
> in a form that a client is able to trust, as part of data exchanged over an 
> HTTPS bundle that already exists between them.
>
> The web server already knows what DNS queries could be triggered by the 
> objects the sever sends the client. For example, they are implied by names 
> embedded in HTML sent from the server to the client. If the client looks up 
> those names using a DNS resolver, that provides opportunites for those 
> queries to be observed by new actors (DNS server operators).
>
> The client of the web server can decide to block or ignore particular objects 
> or parts of objects received by the web server and not use (or look up) the 
> names that relate to things that were blocked. Whether or not DNS queries are 
> made by such a client does not change the information available to the web 
> server operator.

As said the client is the browser and something not easy to change for a lot of 
users. The DNS resolver they use is something users can change, hence I’m not 
concerned about sending queries to it. So let me try to build a simple case of 
what I mean.

The user in his network has setup a PI-hole that blocks lookups to known 
advertising domains with NXDomain hence making tracking with these domains 
impossible.

Now if a browser implements what you purpose it will get the lookup results of 
the IP address of these tracking sites directly embedded in the HTTP 
conversation and immediately connect to them without asking the resolver hence 
circumventing the ad blocking resolver.

So long
-Ralf
——-
Ralf Weber

_______________________________________________
DNSOP mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to