Re: Straw Poll: Restore Header Table and Static Table Indices

On 7 October 2014 12:47, Shigeki Ohtsu <ohtsu@iij.ad.jp> wrote:

> If we reduce a number of static header entries, it would be have a more
> chance for
> a dynamic header to be referred in the index less than 63.
>

That could be a good compromise.  Seeing that three appears to be some
consensus to tweak the table a bit (add static values), then I'd certainly
consider dropping a few of the least frequent static entries to free up
some 1 byte encoding slots for some dynamic entries.

Is the original data used for the frequency analysis still available?   I'm
happy to extract the candidates to be dropped (and suggested values) from
the test data, but it is not that large and I don't know how representative
it is.

If somebody can make the data available, I'm happy to generate a pull
request for a static table with more values and a few less entries.

cheers

-- 
Greg Wilkins <gregw@intalio.com>
http://eclipse.org/jetty HTTP, SPDY, Websocket server and client that scales
http://www.webtide.com  advice and support for jetty and cometd.