Big changes in -17

[Martin Thomson <[e-mail]>]

No, that doesn't mean that we're going to be making major changes, but
I did want to draw attention to a few things that seem likely to
happen in -17 in the transport.

I hope that nothing here is a surprise to anyone.

1. Version negotiation.  This was discussed in NY and people seemed to
like the proposed direction.  There is a PR in progress here:
https://github.com/quicwg/base-drafts/pull/1901

2. MAX_STREAMS.  We realized that the accounting for stream limits,
particularly STREAM_ID_BLOCKED, was slightly broken.  Making the frame
consistent with the transport parameter - that is, having a count
rather than a maximum - solves this problem nicely.  PR:
https://github.com/quicwg/base-drafts/pull/1900

3. Renumbering.  With all the changes we've had, we've tried to keep
identifiers stable.  That means we have a fragmented mess.  The
editors plan to renumber frame types and transport parameters to try
to restore some semblance of sanity.

4. The first octet.  We discussed this in NY:
https://github.com/quicwg/wg-materials/blob/master/interim-18-09/minutes.md#first-octet---martin-thomson
and we will have a decision about the spin bit within 2 weeks.  After
that, one of the designs discussed will be implemented.

5. Your last minute changes.  Maybe.  Get them in soon, please.

After that, the list of remaining design issues gets very lean (at
least for the transport draft).  Once we ship -17, the intent is to
raise the bar considerably for changes that aren't clearly bug fixes
or editorial improvements.  The idea is that people can start
deploying -17, so we can all start to learn the hard lessons.
Avoiding gratuitous changes from that point is going to be necessary.