W3C home > Mailing lists > Public > ietf-http-wg@w3.org > April to June 2013

Re: Updated I-D..

From: James M Snell <jasnell@gmail.com>
Date: Tue, 23 Apr 2013 11:33:43 -0700
Message-ID: <CABP7RbfR0itNxTcOo7+TzfkOKtjEpig1grkjviu6pEbM=tY7HA@mail.gmail.com>
To: "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>
Another rev of the bohe draft addressing a range of items:

  http://www.ietf.org/id/draft-snell-httpbis-bohe-07.txt

* I've added discussion on alternative timestamp and variable-length
integer encodings to the appendix, including some details on why I
chose the options I have. The summary is: when it comes to variable
length integer encoding, there are plenty of generally equivalent
options available. Each encode the same basic information in
approximately the same number of bytes. It really isn't going to
matter much which one we choose. Same thing goes for encoding dates.
I've gone with the Just Pick One approach here.

* Based on feedback I've received off-list and based on testing, I've
decided to change the timestamp value encoding back to millisecond
precision. At worst, it adds only a single byte to our encoding which
is still a significant savings relative to the HTTP-date format.
Millisecond precision delivers new functionality without additional
cost. Implementations that wish to only produce second-level precision
can continue to do so easily.

* I've added discussion in the appendix on how existing standard
header fields can be updated to support the typed value codecs.

* I've added discussion in the appendix on how the new header encoding
scheme can be used to make significant improvements on how the
Set-Cookie and Cookie mechanism works.

* Various spec bugs have been corrected.

As always, feedback is requested and very welcome...

- James


On Sat, Apr 20, 2013 at 10:57 AM, James M Snell <jasnell@gmail.com> wrote:
> All, I've posted an update to the bohe I-D that describes the current
> state of the header encoding options I'm exploring.
>
> http://www.ietf.org/id/draft-snell-httpbis-bohe-04.txt
>
> This is essentially what I described last week with a few tweaks here
> and there. It combines what I feel are the best elements from the
> Delta and HeaderDiff options and rolls in the typed value codecs. I
> have an implementation in github that can be tested with the
> compression-header test suite.
>
> https://github.com/jasnell/http2
>
> As always, comments are welcome.
>
> - James
Received on Tuesday, 23 April 2013 18:34:29 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 11:11:12 UTC