[dnsoverhttp] HTTP/2

[Martin Thomson <[e-mail]>]

I've said this elsewhere, but I don't think that mandating HTTP/2 is
necessary [11].

The draft does this to pick up the security requirements of HTTP/2,
but that can be achieved in other ways that are less disruptive to the
stack.  I would prefer that we soften the requirement and encourage
the use of good practices - HTTP/2 certainly has many virtues that
make it a good choice - but not mandate a specific version.  See also
[56] and [7525].

I do think that the draft needs to be more precise about server push.
Right now, the description of server push is rather oblique and it
could use some more clarity [12].

[11] https://github.com/paulehoffman/draft-ietf-doh-dns-over-https/issues/11
[56] https://mnot.github.io/I-D/bcp56bis/#specifying-the-use-of-http
[7525] https://tools.ietf.org/html/rfc7525
[12] https://github.com/paulehoffman/draft-ietf-doh-dns-over-https/issues/12