Clients use the same crypto handshake after Retry #2746
Conversation
As agreed in London and Tokyo. Closes #2180.
martinthomson
added
design
| first Initial packet. A client can either reuse the cryptographic handshake | ||
| message or construct a new one at its discretion. | ||
| first Initial packet. A client MUST use the same cryptographic handshake | ||
| message it includes in this packet. A server MAY treat a packet that |
There was a problem hiding this comment.
And it MUST NOT reset the packet number to 0.
This isn't just for 0-RTT.
|
We already have an editorial issue in this section; it needs breaking into subsections rather than combining long paragraphs into even longer ones. Thanks for catching the packet number thing. |
| additional 0-RTT packets without constructing a new cryptographic handshake | ||
| message MUST NOT reset the packet number to 0 after a Retry packet; see | ||
| {{packet-0rtt}}. | ||
| packets to the connection ID provided by the server. A client MUST NOT change |
There was a problem hiding this comment.
| packets to the connection ID provided by the server. A client MUST NOT change | |
| packets with the connection ID provided by the server. A client MUST NOT change |
| first Initial packet. A client can either reuse the cryptographic handshake | ||
| message or construct a new one at its discretion. | ||
| first Initial packet. A client MUST use the same cryptographic handshake | ||
| message it includes in this packet. A server MAY treat a packet that |
draft-ietf-quic-transport.md
Outdated
| packets to the connection ID provided by the server. A client MUST NOT change | ||
| the cryptographic handshake message it sends in response to receiving a Retry. | ||
|
|
||
| A client MUST NOT reset the packet number to 0 for any packet number space after |
There was a problem hiding this comment.
Do we need to say "to 0"?
I think it'd be better to just omit "to 0", because resetting the packet number does not necessary mean that it goes to zero. It is totally reasonable for an endpoint to start it's packet number from one.
There was a problem hiding this comment.
It says that you must not reset to zero. It does not matter if you do not necessarily reset to 0 if you sometimes can. I'm not sure why 0 is important in higher packer number spaces, but if it is, it must be stated.
martinthomson
added
editorial
As agreed in London and Tokyo.
Closes #2180.