W3C home > Mailing lists > Public > ietf-http-wg@w3.org > January to March 2016

Re: Submitted new I-D: Cache Digests for HTTP/2

From: Cory Benfield <cory@lukasa.co.uk>
Date: Thu, 14 Jan 2016 11:12:53 +0000
Cc: Kazuho Oku <kazuhooku@gmail.com>, Amos Jeffries <squid3@treenet.co.nz>, HTTP Working Group <ietf-http-wg@w3.org>
Message-Id: <951CA6F0-BED8-4416-8800-8A43B451CEC0@lukasa.co.uk>
To: Stefan Eissing <stefan.eissing@greenbytes.de>

> On 13 Jan 2016, at 09:40, Stefan Eissing <stefan.eissing@greenbytes.de> wrote:
> 
> ...
> Cache-Digest: <base64url encoded, golombset compressed digests>
> Connection: Cache-Digest
> ...
> 
> The question is what a H2 origin server does with such a header,
> should it appear. Ignore, discard?

According to RFC 7540 a H2 origin server would have to reject that stream with RST_STREAM(PROTOCOL_ERROR): see Section 8.1.2.2 for the normative MUST NOT.

For that reason, I think this is a reasonable extension but *only* if the header is defined as connection-specific. Assuming that Kazuho’s objections relating to XMLHttpRequest and FetchAPI are resolved, this approach seems reasonable.

One approach to resolving those problems would be to add a requirement that H2 endpoints treat a stream with the Cache-Digest header in it as a stream error with some useful error flag. That requirement, well policed by origin servers, would probably resolve the issue neatly.

Cory

Received on Thursday, 14 January 2016 11:13:25 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 22 March 2016 12:47:10 UTC