Re: Cache control in trailers?

On 6/02/21 3:34 pm, Mark Nottingham wrote:
> Hey Greg,
> 
> My .02 -
> 
>> On 5 Feb 2021, at 7:16 pm, Greg Wilkins wrote:
>>
>> So Mark... your original proposal was just about the Cache-Control field.   Do you still think that it is only that field that needs support in trailers?
> 
> I never said it was the *only* field that needs support in trailers. However, I believe it's most realistic to focus on a defined use case that has limited impact on implementations, rather than trying to boil the ocean of moving buffering between different parties in the protocol.
> 
> 
>>   What about Set-Cookie, Last-Modified, ETag, Retry-After, Age, Expires, etc.   aren't these fields also affected by the use-case you have outlined?   Do you still think a single field proposal is sufficient or is a more general solution needed?
> 
> I think the property that PHK pointed out -- giving more flexibility for tracking -- is a great argument against doing anything to make Set-Cookie more useful.
> 
> As was already pointed out, ETag can be carried in a trailer today; it just remains for implementations to support it. I suppose we could update Core so that Last-Modified can be too, but I don't think the use case is as compelling.
> 

With my web-dev hat on it is quite common for CMS controlled sites to 
have a huge mix of modification times on scripts and content sources. It 
is a huge pain to calculate a reliable L-M before output in this 
environment. It would much cleaner to (be able to) accumulate a 
modification time across the scripts and dump a Trailer header with the 
true L-M value.


> Retry-After isn't terribly useful for 99%+ of the web, and isn't supported in browsers AFAIK, so it's not very interesting.
> 
> Updating Age doesn't make sense; you need it to be as reliable as possible, and adding doubt about that is just going to make things worse.
> 

Ack. No need for Age at all on freshly generated content use-cases.


> Expires - *shrug* I guess. I'd rather kill it in favour of CC: max-age -- again, we should think about what behaviours we're encouraging.
> 

Very much agreed on deprecating Expect. Though the value limit on 
max-age is smaller than possible Expect values. So there is a use-case 
for extremely long duration archiving to use Expect.


Amos

Received on Sunday, 7 February 2021 09:03:24 UTC