Re: Suggestion for simplified HTTP 2.0 header compression algorithm


Thank you all for your comments.

You are right Amos, the static headers are not re-transmitted constantly.  They are only sent in the SETTINGS frame.

The following text is probably  a bit misleading:

> Static headers and dynamic headers are combined in the following way:


> 1.Emit the dynamic headers.


> 2.Emit the static headers whose header names are not present in the

> dynamic headers.

We tried to describe how the client/server combines the static and dynamic headers into one header set for the request, when it receives a HEAD frame with dynamic headers.

We were not aware of a fact that there is another draft ( in the same area.

Our proposal was inspired by the way how static and dynamic headers are exchanged in the WAP WSP protocol.

The header encoding in is better than the one which we propose.

We should use it.

Regarding your comments Michael - you are right that the solution with sending static headers in the SETTINGS frame doesn't work well for proxies. However we believe that the current header compression scheme will most likely also not achieve high compression ratio in that case (Multiplexing requests from many "UE" on one connection).

We would recommend not to send any static headers in the SETTINGS frame in proxy case if requests from multiple UE's are multiplexed on connection.

Regarding Marks comment - no we do not have any running code yet, we wanted to see the feedback on our proposal before development. However, we do have a running implementation of current HPACK proposal (draft v 06). When we implemented that one, we came to the conclusion that it is too complicated and we fear that it will cause interworking problems in the future.

Best regards,

Hans Spaak
Ericsson AB

Received on Thursday, 7 November 2013 14:31:28 UTC