W3C home > Mailing lists > Public > ietf-http-wg@w3.org > January to March 2019

Re: Cache Digests status

From: Lucas Pardue <lucaspardue.24.7@gmail.com>
Date: Thu, 24 Jan 2019 11:51:44 +0000
Message-ID: <CALGR9obko4PiQY9oEJj0BUE-WJTwGGnzbavcOKN5py-nMYpWYg@mail.gmail.com>
To: Sebastiaan Deckers <sebdeckers83@gmail.com>
Cc: Loïc Hoguin <essen@ninenines.eu>, Mark Nottingham <mnot@mnot.net>, "ietf-http-wg@w3.org Group" <ietf-http-wg@w3.org>, Patrick McManus <mcmanus@ducksong.com>, Tommy Pauly <tpauly@apple.com>, Kazuho Oku <kazuhooku@gmail.com>, Yoav Weiss <yoav@yoav.ws>
On Thu, 24 Jan 2019, 11:35 Sebastiaan Deckers <sebdeckers83@gmail.com wrote:

> Since Node.js does not expose a low level API to HTTP/2 frames it is
> impossible (afaik?) to implement the spec properly at this time. Not sure
> about other popular frameworks/languages and their HTTP/2 implementations.
> Perhaps affordance could be made in the spec to use headers instead of
> custom frames. Or maybe implementors of HTTP/2 could be encouraged to
> expose frame-level access, just like custom HTTP header fields

I think this is an important observation. The utility of H2 extension
frames is restricted by their ease of use (or lack thereof). It is very
difficult to experiment with ideas for extension frames. The current
landscape seems to place the client-side as a gate with a high bar. In
contrast, headers are very easy to experiment with and build a supporting

If we consider the origins of Cache Digest, would it have even been adopted
without the evidence from the sub-optimal cookie-based prior work?

Frames have some unique characteristics that are not possible with headers
alone.  It is a shame that the present state of affairs make it hard to
realize their potential.

Received on Thursday, 24 January 2019 11:52:17 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 24 January 2019 11:52:18 UTC