I have reviewed this document as part of the Operational directorate's ongoing effort to review all IETF documents being processed by the IESG.  These comments were written with the intent of improving the operational aspects of the IETF drafts. Comments that are not addressed in last call may be included in AD reviews during the IESG review.  Document editors and WG chairs should treat these comments just like any other last call comments.

This document describes a version negotiation mechanism that allows a client and server to
select a mutually supported version. The mechanism can be further broke down into incompatible Version Negotiation
and compatible Version Negotiation.

I believe this document is on the right track and but worth further polished with the following editorial comments.
Major Issues:
No
Minor Issues: 
1. Section 4 introduce verson downgrade term
[Qin]What the version downgrade is? I feel confused, does it mean when I currently use new version,
I should not fall back to the old version?
Can you explain how version downgrade is different from version incompatible?
It will be great to give a definition of version downgrade in the first place or provide
an example to explain it?
2. Section 9 said:
" The requirement that versions not be assumed compatible mitigates the possibility of cross-protocol attacks, but more analysis is still needed here."
[Qin] It looks this paragraph is incomplete, what else analysis should we provide to make this paragraph complete?
3. Section 10
[Qin]: I am not clear why not request permanent allocation of a codepoint in the 0-63 range directly in this document.
In my understanding, provisional codepoint can be requested without any dependent document? e.g., Each vendor can request IANA for Vendor specific range.
Secondly, if replacing provisional codepoint with permanent allocation, requires make bis for this document, I don't think it is reasonable.
Nits:
1. Section 2 said:
"
For instance, if the client initiates a connection with version A and the server starts incompatible version
negotiation and the client then initiates a new connection with ....
"
[Qin]Can the client starts incompatible version negotiation? if not, 
I think we should call this out, e.g., using RFC2119 language.
2. Section 2, last paragraph
[Qin] This paragraph is a little bit vague to me, how do we define not fully support?
e.g., the server can interpret version A, B,C, but the server only fully implements version A,B,
regarding version C, the server can read this version, but can not use this version, in other words,
it is partially implemented, is my understanding correct?
3.Section 2.1 the new term "offered version"
[Qin] Can you add one definition of offered versions in section 1.2, terminology section?
To be honest, I still not clear how offered version is different from negotiated version?
Also I suggest to add definitions of accepted version, deployed version as well in section 1.2?
Too many terminologies, hard to track them in the first place. 
4. Section 6 said:
"
it is possible for some application layer protocols to not be able to run over some of the offered compatible versions. 
"
[Qin]I believe compatible versions is not pertaining to any application layer protocol, if yes,
 s/compatible versions/compatible QUIC versions
5.Section 7.1 said:
"For example, that could be accomplished by having the server send a Retry packet in the original
 version first thereby validating the client's IP address before"
[Qin] Is Version first Version 1? If the answer is yes, please be consistent and keep using 
 either of them instead of inter-exchange to use them.
 s/version first/version 1