W3C home > Mailing lists > Public > ietf-http-wg@w3.org > July to September 2021

Re: #1605: Targeted CC as a trailer?

From: Mark Nottingham <mnot@mnot.net>
Date: Fri, 27 Aug 2021 12:16:31 +1000
Cc: ietf-http-wg@w3.org
Message-Id: <737F55E0-B1C4-4D2D-B277-E262C7914FAF@mnot.net>
To: Martin Thomson <mt@lowentropy.net>


> On 27 Aug 2021, at 12:12 pm, Martin Thomson <mt@lowentropy.net> wrote:
> 
> Isn't the primary advantage of x-cache-control that it behaves just like cache-control?  This would create inconsistencies that would make it harder to adapt to different values of "x".  I would prefer to see a general solution.

This doesn't change the behaviour so much as it adds to it, I think. The semantics of the cache directives would be the same, as would the syntax of the field; it only has additional semantics when it occurs in a trailer.

(I somewhat suspect that there isn't going to be immediate interest in this one, which means we're likely to punt it, but I thought I'd ask and perhaps be surprised)


> On Fri, Aug 27, 2021, at 12:06, Mark Nottingham wrote:
>> <https://github.com/httpwg/http-extensions/issues/1605>
>> 
>> We recently had a fairly long discussion about whether we could define 
>> Cache-Control as a trailer, so that caching policy could be updated 
>> after content is sent. We concluded that there wasn't yet enough 
>> interest or alignment to pursue specification work there.
>> 
>> I've raised this issue for Targeted Cache-Control to see if there's any 
>> interest in defining it for these headers. For example, a CDN or 
>> reverse proxy could support trailer updates of caching policy when 
>> targeted specifically at them. 
>> 
>> From a specification standpoint, we could do that by explaining how it 
>> would work, but requiring the specific header field definition to 
>> explicitly opt into being valid in trailers. I'd say CDN-Cache-Control 
>> wouldn't do this, but other fields could.
>> 
>> Alternatively, we could leave the spec as-is, and allow individual 
>> headers to specify how they do this. The potential issue there is that 
>> they might divert in doing so.
>> 
>> Are folks (especially those who have a cache implementation) interested in this?
>> 
>> Cheers,
>> 
>> --
>> Mark Nottingham   https://www.mnot.net/
>> 
>> 
>> 
> 

--
Mark Nottingham   https://www.mnot.net/
Received on Friday, 27 August 2021 02:16:49 UTC

This archive was generated by hypermail 2.4.0 : Friday, 27 August 2021 02:16:50 UTC