Ted, This would be great. However, though I'm not an expert at all, http-semantics is deeply integrated into quic-http, and at 200+ pages is reasonably likely to have a small edit here and there. I'm skeptical we can ship HTTP/3 without this anchor firmly lodged.
But I'm happy to be corrected by people with more knowledge of both documents! On Mon, Jan 11, 2021 at 1:04 PM Ted Hardie <[email protected]> wrote: > Hi Martin, > > The easiest way to handle this would be to have those included in the > downref registry. That would allow the documents to proceed with a > reference to the drafts, rather than the final RFCs. If the ADs are > confident that the -chache and -semantics will not change the elements that > are referenced by QUIC, then a second last call that highlights that point > (and basically nothing else) will unstick this. > > As someone has been waiting for cluster 238 for quite a while now, I am a > big fan of not trying to wait until everything can go lockstep. The > benefits have definitely not outweighed the costs for C238, and I don't > think they would here. Downref it and go on with publication. > > Just my opinion, of course. > > regards, > > Ted > > On Mon, Jan 11, 2021 at 12:57 PM Martin Duke <[email protected]> > wrote: > >> I was disappointed to realize that quic-http has two normative references >> (httpbis-cache and https-semantics) that are not even in WGLC. >> >> This suggests that there will be significant delay (best case, months) >> between QUIC and HTTP/3 finalizing as RFCs. >> >> How are we to handle deployment of this? Will there be endpoints >> operating QUICv1 with h3-29 or h3-32? Or have the chairs cooked up a >> maneuver to avoid this problem? >> >> Martin >> >
