Re: [quicwg/base-drafts] Extension frames (#1068)

[Christian Huitema <[e-mail]>]

Now that I read EKR's comment, I believe he might be correct. Or maybe not. Defining the frame type as a varint would remove the need for dynamic mapping. On the other hand, there will still be a need for a negotiation, to check whether the option is actually supported by the peer. The implementation in C would be:

1) With the current mapping definition, use a table of function pointers to map the extension frame type to the type specific coding/processing functions. The mapping table is filled up when processing the TP parameter. Unused values point to some error code.

2) With the varint coding of frame type, add some tests for varint decoding. Use switch statements with fixed values to point to the implementation-supported extensions. Use a table of booleans to indicate which implementation-supported extensions are valid for the connection. The boolean table is filled up when processing the TP parameter. Unsupported values and non negotiated values point to some error code.

This seems pretty similar in term of complexity. I am not enthusiastic about varint decoding of frame types, and I think if would only be needed if some application required more that 128 extension frames simultaneously for a given connection. All things considered, I like the currently proposed design.

-- 
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/pull/1068#issuecomment-378713814