[quicwg/base-drafts] Use a "stream" for transmitting CIDs (#1826)

[Kazuho Oku <[e-mail]>]

Transmission of new and retired CIDs is something that do not need low latency, but it seems like it is going to have lot of code specific to the purpose.

Looking at #1799, I am starting to wonder if we could use a "stream" to carry the new and retiring CIDs (sorry for suggesting to change the approach at a late moment).

**The proposed design**

Like the CRYPTO frame that carries the TLS messages, We will have a CID frame that carries the CID messages. This is essentially a "CID stream" much like the "CRYPTO stream."

The format of a CID frame will be equivalent to the CRYPTO frame.

A CID message has the following structure. Type is either NEW_CID (0) or RETIRE_CID (1).

```
 0                   1                   2                   3
 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|    Type (8)   |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                          Length (i)                         ...
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                       CONNECTION ID (*)                     ...
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
```

Instead of flow-control based on octets, the number of CIDs given to the peer by the use of the NEW_CID messages will be governed by a MAX_NEW_CID frame that is equivalent to the MAX_STREAM_ID frame, except for the fact that it carries the number of CIDs.

**The Upsides**

Code reuse _and_ design reuse.

We can reuse a lot of code. Basically, we only need glue code and codec for the CID messages. All the code that relates to retransmission and back-pressure can be reused. We are also reusing a design that we are familiar with.

These aspects lead to less issues.

**The Downsides**

Head-of-line blocking in the delivery of CIDs. But I do not think that is an issue in practice.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/quicwg/base-drafts/issues/1826