Re: [quicwg/base-drafts] Transport error codes would be smaller and more consistent as a varint (#2672)

[Kazuho Oku <[e-mail]>]

I also think that varint is OK. It gives application protocols to use up to 62-bit space for error codes rather than enforcing them to use 16-bit space. The downside is that the CONNECTION_CLOSE frame could become one byte bigger, but I do not think that is a real cost.

Re HTTP_MALFORMED_FRAME, my preference goes to either a) recommend endpoints expressing the offending frame type in the reason phrase, or b) change the definition of application-close frame (i.e. CONNECTION_CLOSE with type=0x1d) to carry a blob, and define in the HTTP/3 spec. that the blob would be a tuple of error-code (varint), offending-frame-type (varint), and a reason phrase.

My view is that 2^60 is a dirty hack. It works, but I'd argue that it is a sign the HTTP draft doing something unexpected by the transport design. Therefore, I'm arguing to do what is expected by the transport, or change the design of the transport to meet the requirement of HTTP.

-- 
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/2672#issuecomment-490700673