Re: Header Stats

[James M Snell <[e-mail]>]

Ok.. I have updated the calculations to show:

  1. General variability of header values. The lower the number, the more
redundant the value tends to be.
  2. A frequency distribution of specific values per header. This is rather
verbose but extremely informative
  3. For date and numeric header values, comparison values between the
text-value and optimized binary encoding value.
  4. A summation of the total bytes saved by using the optimized binary
encoding for dates and numeric headers.

TODO:

  - Implement experimental Set-Cookie, Cookie and Cache-Control headers to
see the difference for binary encoding
  - Properly handle null-separated value lists

If there are other interesting calculations you'd like to see, let me
know...

The updated output is here:
https://github.com/jasnell/compression-test/tree/master/counts


On Tue, Jan 22, 2013 at 1:24 PM, James M Snell <[e-mail]> wrote:

> I've started working on generating stats for individual headers within
> messages. Rather than take up too much space here on the list for the
> results, I am keeping the results in my github fork [1] of the
> compression-test code and will be posting summaries of the results
> periodically on my personal blog [2]. I will be putting together a summary
> of my findings together in time for the interim meeting next week.
> Unfortunately, however, I will not be able to attend the meeting.
>
> [1] https://github.com/jasnell/compression-test/blob/master/counts/
> [2] http://chmod777self.blogspot.com/2013/01/http-20-header-stats.html
>
> - James
>