Re: HTTP/2 Server Push and solid compression

Matthew Kerwin <matthew@kerwin.net.au> Tue, 21 May 2019 22:48 UTC

Resent-Date: Tue, 21 May 2019 22:47:02 +0000
Resent-Message-Id: <E1hTDXC-0005Wa-9l@frink.w3.org>
MIME-Version: 1.0
References: <CA+phaedE0m4LniC38GBkJ-M0gAph0LSSGhQ1ZWJE6k0UOFcokw@mail.gmail.com> <CA+phaecOMRAd8R=oEYj+DMVkzaVKq5Qbt9AxECrtofLqMxeKQA@mail.gmail.com> <898C6E1B-73ED-4F13-B435-745680E662AB@felipegasper.com> <CA+phaefGgNidhqkaHaX6kq6aEK0n5EV9j_Qya-9hku0CzHN=0Q@mail.gmail.com> <65908183-BFD4-447D-9F03-E4EB9EB6E0EE@felipegasper.com> <20190521180752.GA1102985@LK-Perkele-VII>
In-Reply-To: <20190521180752.GA1102985@LK-Perkele-VII>
From: Matthew Kerwin <matthew@kerwin.net.au>
Date: Wed, 22 May 2019 08:46:26 +1000
Message-ID: <CACweHNACboW3xMQdLn1ZefmkVO8Z6s1bNbYU0m+coZJg4MUx=w@mail.gmail.com>
To: Ilari Liusvaara <ilariliusvaara@welho.com>
Cc: Felipe Gasper <felipe@felipegasper.com>, Alan Egerton <eggyal@gmail.com>, HTTP Working Group <ietf-http-wg@w3.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Received-SPF: pass client-ip=209.85.166.173; envelope-from=phluid61@gmail.com; helo=mail-it1-f173.google.com
Subject: Re: HTTP/2 Server Push and solid compression
Archived-At: <https://www.w3.org/mid/CACweHNACboW3xMQdLn1ZefmkVO8Z6s1bNbYU0m+coZJg4MUx=w@mail.gmail.com>
Resent-From: ietf-http-wg@w3.org
Resent-Sender: ietf-http-wg-request@w3.org
Precedence: list

On Wed, 22 May 2019 at 04:10, Ilari Liusvaara <ilariliusvaara@welho.com> wrote:
>
> On Tue, May 21, 2019 at 01:51:27PM -0400, Felipe Gasper wrote:
> >
> > Possibly … does h2 multiplexing use a separate compression context
> > for each stream, or does it funnel each message through the same
> > context? If the former, then I would think it’s a non-issue since
> > streams are processed sequentially.
>
> AFAIK, HTTP/2 can not compress message bodies. The mechanism HTTP/1.x
> had to compress message bodies (Transfer-Encoding) was removed.
>

Just because it's not in the spec, doesn't mean it can't be in *a* spec. [1]

> And even the HTTP/1.x mechanism used separate context for each
> request or response.
>

Yeah, the argument for TE was more for sub-resource compression
(request a byte range from an identity-encoded resource, like a log
file, and compress the resulting chunk in transit), rather than
super-resource compression.

> And I would imagine reusing contexts would run into all sorts of
> data leaks by abusing compression...
>

That's why it got yanked from TLS, yeah.

Cheers

[1]: https://tools.ietf.org/html/draft-kerwin-http2-encoded-data

-- 
  Matthew Kerwin
  https://matthew.kerwin.net.au/

HTTP/2 Server Push and solid compression Alan Egerton
Re: HTTP/2 Server Push and solid compression Alan Egerton
Re: HTTP/2 Server Push and solid compression Felipe Gasper
Re: HTTP/2 Server Push and solid compression Alan Egerton
Re: HTTP/2 Server Push and solid compression Felipe Gasper
Re: HTTP/2 Server Push and solid compression Patrick McManus
Re: HTTP/2 Server Push and solid compression Ilari Liusvaara
Re: HTTP/2 Server Push and solid compression Alan Egerton
Re: HTTP/2 Server Push and solid compression Matthew Kerwin
Re: HTTP/2 Server Push and solid compression Matthew Kerwin
Re: HTTP/2 Server Push and solid compression Kari Hurtta
Re: HTTP/2 Server Push and solid compression W. Felix Handte
Re: HTTP/2 Server Push and solid compression Kari Hurtta