[quicwg/base-drafts] QPACK: instructions and stream description is suboptimal (#2339)

[Lucas Pardue <[e-mail]>]

Did the section hierarchy get messed up a little in a recent refactor?

The concept of QPACK instructions are first introduced in section 2, along with encoder and decoder streams. There is no forward reference to the later definitions.

The real problem I have is with section 4.2. It contains a bulleted list that is almost identical to section 2, but subtly different. The text following this list is also ambiguous

> There MUST be exactly one of each unidirectional stream type in each direction.
The previous list referred to request streams and push streams. These are obviously out of scope but it could get mistaken.

I think we've discussed elsewhere that implementations could decide not to use dynamic compression, and so the encoder and decoder streams are not necessary. I think this sentence could be made clearer by saying something like "if the dynamic table is being used, each endpoint MUST create exactly one pair of encoder and decoder streams."

> Receipt of a second instance of either stream type MUST be treated as a connection error of HTTP_WRONG_STREAM_COUNT.

Perhaps simplify to "receipt of additional streams which claim to be an encoder or decoder stream MUST be treated as a connection error of HTTP_WRONG_STREAM_COUNT.

> Closure of either unidirectional stream MUST be treated as a connection error of type HTTP_CLOSED_CRITICAL_STREAM.

Perhaps "The closure of encoder or decoder streams, at any point, MUST be treated as a connection error of type HTTP_CLOSED_CRITICAL_STREAM."

> This section describes the instructions which are possible on each stream type.
No it doesn't :)

While I am here, Section 4.5 is titled "Request and Push Streams" but it just talks about instructions to include in frames. 

I'm left with a feeling that there is a misplaced focus on streams rather than instructions. This could be inverted to state instructions, with a comment on where they are allowed to appear, summarised in a nice table somewhere in the document.





-- 
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/2339