Re: Rechartering HTTPbis

On 28/01/2012 1:40 p.m., Poul-Henning Kamp wrote:
> In message<EC463243-A3B0-4E35-BE92-95B35F3DE067@mnot.net>, Mark Nottingham wri
> tes:
>
>>> Re HTTP/next: it would be good to collect a list of things we think we =
>> should make progress on; not surprisingly, I'd nominate I18N for header =
>> field values.
>>
>> So, that's an interesting question.
> I18N in transport headers:  Forget it, it would kill performance if
> done right, and do nobody any good if not.
>
> I18N in content headers: Sure, be my guest.
>
>

Skipping over I18N and its multiple charsets and going specifically to 
UTF-8 representation of the existing header could be an option though.

We already have a set of well defined transport headers with defined 
ASCII/UTF-8 interchangeable characters. New headers will have to 
consider whether they exist only on HTTP/2 links or also apply to HTTP/1 
links and choose relevant naming characters based on that. Transport 
being hop-by-hop this is a choice they can make safely.

The message and entity meta headers which cross hops  is where the 
problems occur by the bucket full. I somehow doubt it would be a good 
idea to make two distinct header formats, but who knows.

AYJ

Received on Saturday, 28 January 2012 02:08:51 UTC