Hacker News Reader

Comment by ivanr | Hacker News Reader

Take a look instead at the DNS SVCB and HTTPS resource records, which have already been adopted in practice:

- Service binding and parameter specification via the DNS (DNS SVCB and HTTPS RRs) https://datatracker.ietf.org/doc/draft-ietf-dnsop-svcb-https...

View on HN

It appears Google implemented the records type65, but not type64. Also Youtube. I am curious what browsers are leveraging this currently. I can't find many other domains using it yet. It seems Apple use it on some subdomains [1]

    dig +noall +answer -t type65 google.com
    google.com.  2857 IN HTTPS 1 . alpn="h2,h3"

I see some hits in my name server stats for https/svcb types so I guess bots are looking for it. Perhaps scanning how widely adopted? The list of types has grown since I last looked. [2] People have been busy.

[1] - https://serverfault.com/questions/1075522/whats-the-use-case...

[2] - https://en.wikipedia.org/wiki/List_of_DNS_record_types

Chrome at least, behind a flag.[1]

[1] https://chromestatus.com/feature/5485544526053376

Beat me to it, I was also wondering if the OP proposal hasn't been completely obsoleted by SVCB/HTTPS at this point.

Still reveals a sad picture of the ecosystem that a standard which addressed the same shortcomings existed since 2009 but was completely ignored until Cloudflare reinvented it in a proprietary fashion.

On the other hand, expecting that end users suddenly start typing "https+srv://" everywhere for exactly no added benefit except to make standards wonks happy, doesn't exactly inspire much confidence in the standard's authors either...

The -00 draft[1] did not require the +srv. I would guess the blame lies with security problems when user agents with support for the SRV record see one thing and those without see another.

(The drafts go up to -05, though this is not apparent from the posted link.)

[1] https://datatracker.ietf.org/doc/html/draft-jennings-http-sr...

Does it support specifying the HTTPS port? This is a very important use case for selfhosting. I skimmed through Cloudflare's article[0] but didn't see any mention of it.

[0]: https://blog.cloudflare.com/speeding-up-https-and-http-3-neg...

Ok after a little searching the spec itself talks about a port parameter in section 7.2. So it would appear SVCB can indeed replace HTTP SRV for my purposes.

Does this deprecate TTLS records? I've only just begun to generate those

Reply on news.ycombinator.com