Re: Last Call: <draft-yevstifeyev-http-headers-not-recognized-08.txt> ('Headers-Not-Recognized' HTTP Header Field) to Experimental RFC

[Mykyta Yevstifeyev <[e-mail]>]

17.12.2010 16:14, Daniel Stenberg wrote:
> On Fri, 17 Dec 2010, Mykyta Yevstifeyev wrote:
>
>>> What do packets have to do with HTTP headers?
>
>> What do you mean? Packets have nothing to do with headers, there is 
>> nothing about this in paragraph above. Maybe you meant middle-boxes?
>
> Read through your -09 spec again. You speak of "HTTP packets" in 
> several places, where they should rather be "HTTP request" or "HTTP 
> responses" etc.
Maybe I meant HTTP messages here.
>
> I also find the use of 'host' very confusing in the document as it is 
> clearly used instead of the more proper 'client' or 'server' in some 
> places.
In previous version there have been the 'server' and 'client' terms 
instead 'host'. However it is obvious for me that there can be as 
servers as clients that do not recognize some headers of another side of 
exchange.
>
> The second paragraph in section 2.1 combines both these mistakes and 
> make a blob of text that I cannot understand:
>
>    When HTTP host receives HTTP packet with Headers-Not-Recognized
>    header, it is RECOMMENDED that it avoids sending packets with headers
>    with mentioned in it names or tries to change them so that it is able
>    to recognize and process them.
>
> Does this say that if a client learns about headers that the server 
> doesn't support, it shouldn't send them in subsequent requests?
No, that just describes the behavior of the host after receiving the 
H-N-R header. However this paragraph needs improvement, and I'll do that 
in the next version of the draft, which will bw available at the end of 
Last Call.

All the best,
Mykyta Yevstifeyev