2023年10月13日(金) 9:18 Martin Thomson <m...@lowentropy.net>: > On Fri, Oct 13, 2023, at 11:11, Kazuho Oku wrote: > > 2023年10月11日(水) 10:07 Martin Thomson <m...@lowentropy.net>: > >> I'm more concerned about adding another negotiation point for the > request. If they progress in parallel, then we need separate negotiation > for each and that gets tricky when there is an interdependency like this. > > > > If that is the main concern, I think we might consider changing the > > Transport Parameter used for negotiating Reliable Reset to take a > > version number. > > I'm not enthusiastic about adding extra extension points like that. > > My sense is that either ENOUGH is worth doing or not, and that won't > change much in future. If the protocol needs to get much more complex in > order to support it, then it will be less worthwhile. >
Thanks, I think I'm fine with that too. After all, it is not difficult to have something like ENOUGH inside the application protocol, as such a signal can be sent on a control stream. That's very different from CLOSE_STREAM, without which it is complicated to guarantee partial delivery of stream data. -- Kazuho Oku
