Re: Call for Adoption: Cache Digests for HTTP/2

FTR: Apache httpd has an implementation of version 0 of the draft, using if from Headers and also to keep internal state of the connection, done by me.

Unsurprisingly, I support adoption of the draft. I had only a glance over of the current changes, there are some interesting additions. Ultimately, feedback from client/intermediaries implementors are necessary here. And I think this WG is a good place to make that happen.

-Stefan

> Am 22.06.2016 um 09:26 schrieb Cory Benfield <cory@lukasa.co.uk>:
> 
> 
>> On 22 Jun 2016, at 07:52, Mark Nottingham <mnot@mnot.net> wrote:
>> 
>> Please state whether you support adoption, and ideally why. Expressions of interest in implementation would also be very helpful.
> 
> I support adoption of this draft. I will also commit to building a partial implementation: that is, I will add support for handling and emitting the CACHE_DIGEST frame to my Python libraries, but I will not commit to updating any Python server or client to respecting or emitting CACHE_DIGEST by default.
> 
> The Cache Digest draft strikes me as an interesting way to resolve a very real tension in HTTP/2. HTTP/2 has a serious focus on reducing the number of network round-trips that are needed for a client to obtain all the resources, but it provides no obvious way for a client to inform a server or intermediary that certain resources are not required. RST_STREAM, as this draft points out, is usable in this role, but certainly not ideal.
> 
> For me, then, Cache Digest seems like a useful extension to HTTP/2. It’s certainly not mandatory, and its existence doesn’t hint at an oversight in RFC 7540, but it’s the perfect example of what a HTTP/2 extension can do: provide better information sharing between the client and the server to improve their efficiency.
> 
> Cory

Received on Thursday, 23 June 2016 08:45:04 UTC