Re: #90: multipart/byteranges

I think the issue is that allowing a special case where it is allowed  
makes parsers significantly more complex, and isn't in wide use anyway.


On 14/11/2008, at 1:49 PM, Henrik Nordstrom wrote:

> On tor, 2008-11-13 at 17:10 -0800, Mark Nottingham wrote:
>> So far, discussion of this issue <http://trac.tools.ietf.org/wg/httpbis/trac/ticket/90
>>> seems to centre around disallowing the use of multipart/byteranges
>> as a message delimiter, because it is not widely implemented, and
>> because chunked encoding can be used instead.
>>
>> Any objections to doing so?
>
> None here, except that I don't really see it needed.
>
> P1 4.4 Message Length (and 2616 4.4) already supports this view
> indirectly, with multipart/byteranges being the next lowest priority
> message delimiting, only close of connection has lower priority..
>
> P1 4.4 #4 also considerably restricts when this media type is allowed
> with a clear MUST NOT.
>
> The text in the multipart/byteranges appendix is only relevant in the
> scope where this media type is allowed to be used. P5 Appendix A.
> Internet Media Type multipart/byteranges is also quite clear on that
> this media type only applies to 206 responses under specific
> conditions.
>
> So I can't find anything even remotely indicating multipart/byteranges
> is meant to be generally accepted or parsed outside the context of 206
> responses to multi-range request.
>
> Regards
> Henrik


--
Mark Nottingham     http://www.mnot.net/

Received on Saturday, 15 November 2008 02:49:47 UTC