Re: Huffman [was: Re: h2 header field names]

On 9/5/14, 2:10 PM, "Poul-Henning Kamp" <phk@phk.freebsd.dk> wrote:


>--------
>In message 
><CABkgnnUnoSQnpPSsi4YdEz2DhPsF90WAADhR-imeKU2WDwuqNA@mail.gmail.com>
>, Martin Thomson writes:
>>On 5 September 2014 10:05, Poul-Henning Kamp <phk@phk.freebsd.dk> wrote:
>>> It doesn't answer my question if HTTP/1 headers which cannot ever
>>> be sent through HTTP/2 were filtered out before estimating the PDFs.
>>
>>https://code.google.com/p/chromium/codesearch#chromium/src/net/spdy/spdy_
>>http_utils.cc&sq=package:chromium&rcl=1409711490&l=109
>>
>>...indicates that those were removed.
>
>How is that code related to the data that went into the huffman table ?
>
>Also: Entries from the static table should also be removed.

And all (remaining) header field names should be converted to lowercase.
This appears to have occurred in the code above.

Were all of these aspects documented anywhere or is this code the only
documentation we have?  I agree with PHK - we are likely to see folks
wishing to understand the process and I believe transparency and clear
documentation are the best answer.

Received on Friday, 5 September 2014 22:46:31 UTC