Re: #541: CONTINUATION

Not an implementer of HTTP myself - those are in the other building, but I do meet with them regularly (and no, we won’t have something ready in the relevant timeframe)

I don’t believe that the interop and deployment experience from draft-13 will provide new information for this issue. 

I’d also like to point out that #2 has an issue with middleboxes. Middleboxes futz with headers. So a client or server may send headers that are less than 16K when HPACK’d, but after futzing and re-HPACK-ing, those can exceed 16K. So a client sends a HEADERS frame that seems entirely valid, only for the stream to be terminated by the proxy because of the excessive size of the headers.

I would have been fine with jumbo-frames even if they’re only for headers (although I’d like them to be for everything), even if they were a negotiated option. But absent jumbo-frames on the menu, I’d go with #0.

Yoav


On Jul 2, 2014, at 2:12 PM, Mark Nottingham <mnot@mnot.net> wrote:

> 
> 0) the status quo
> 
> 1) <https://github.com/http2/http2-spec/pull/544> and variants thereof (e.g., not including CONTINUATION in flow control; alternative syntaxes)
> 
> 2) limiting header sizes to 16K (HPACK’d) in HTTP/2, as per PHK’s suggestion
> 
> There’s also another implicit option; 
> 
> 3) We can’t know until we get substantial interop and deployment experience with draft-13.
> 

Received on Wednesday, 2 July 2014 13:57:37 UTC